Brandon Lazovic writes about how AI is changing the way people find and buy products, and what merchants actually control as agents and answer engines take over.https://brandonlazovic.dev/https://brandonlazovic.dev/pulse/2026-07-31/#gemini-spark-chrome-auto-browsehttps://brandonlazovic.dev/pulse/2026-07-31/#gemini-spark-chrome-auto-browseGoogle rolled out Chrome integration for its Gemini Spark assistant on July 30, 2026, adding an auto-browse mode that can use a user's saved accounts and passwords to handle errands such as scheduling apartment viewings, researching flights, and starting a booking. The feature starts in the US for Google AI Pro subscribers, alongside separate access expanding to more than 160 additional countries, and hands control back to the user before payments. Why it matters: A production browser agent that logs into real accounts and completes bookings puts a site's semantic structure and checkout flow directly in the path of autonomous task completion, not just search visibility.Fri, 31 Jul 2026 00:00:00 GMThttps://brandonlazovic.dev/pulse/2026-07-31/#gpt-5-6-luna-terra-price-cutshttps://brandonlazovic.dev/pulse/2026-07-31/#gpt-5-6-luna-terra-price-cutsOpenAI said on July 30, 2026 that GPT-5.6 Luna, its fastest and most affordable model, will cost 80% less, and GPT-5.6 Terra, its balanced everyday model, will cost 20% less, effective immediately. The cuts also lower how usage counts against Codex and ChatGPT Work subscriptions. Separately, a new Fast mode for GPT-5.6 Sol in the API runs up to 2.5x faster than Standard processing at twice the price, replacing Priority Processing entirely. Why it matters: A model this much cheaper at the volume tier changes the math on which classification and extraction workloads are worth automating at scale for an SEO or e-commerce data pipeline.Fri, 31 Jul 2026 00:00:00 GMThttps://brandonlazovic.dev/pulse/2026-07-31/#avatarin-gpt-realtime-retail-agenthttps://brandonlazovic.dev/pulse/2026-07-31/#avatarin-gpt-realtime-retail-agentOpenAI published a case study on July 30, 2026 describing how avatarin, an AI customer-service company spun out of ANA Holdings, built a 24/7 multilingual voice shopping agent for Japanese retailer Yamada Denki using GPT-Realtime. In a two-week public campaign on Yamada Denki's online store, roughly 30,000 shoppers used the agent, and 92% of post-use survey responses were positive. A retrieval-augmented system grounds its answers in live product data. Why it matters: A named retailer running a voice agent past tens of thousands of real shoppers with a measured satisfaction rate is a harder data point for agentic commerce than another vendor demo.Fri, 31 Jul 2026 00:00:00 GMThttps://brandonlazovic.dev/pulse/2026-07-31/#wp-engine-bigcommerce-commerce-connecthttps://brandonlazovic.dev/pulse/2026-07-31/#wp-engine-bigcommerce-commerce-connectWP Engine announced Commerce Connect on July 29, 2026, a partnership with BigCommerce's parent company Commerce that lets WordPress-based stores add enterprise commerce capabilities without a platform migration. The companies say existing themes, designs, and URL structures stay intact, and sites remain operational during the upgrade, aiming to avoid the traffic and search-visibility loss common in a full replatform. The offer targets high-growth, mid-market brands and agencies managing expanding catalogs and rising traffic. Why it matters: A no-replatform path to enterprise commerce features removes one of the biggest reasons growing merchants have historically had to accept SEO risk just to scale their checkout.Fri, 31 Jul 2026 00:00:00 GMThttps://brandonlazovic.dev/pulse/2026-07-31/#google-search-box-pages-quality-issuehttps://brandonlazovic.dev/pulse/2026-07-31/#google-search-box-pages-quality-issueOn the July 30, 2026 episode of Search Off the Record, Google's John Mueller and Martin Splitt said spammers exploit a site's internal search function to generate indexable pages stuffed with pharmaceutical, adult, or casino keywords, piggybacking on the domain's authority. Google can flag the affected site the same way it flags hacked content in Search Console, and Mueller called noindex the cleaner fix, since robots.txt only blocks crawling and a disallowed URL could still theoretically appear in the index. Why it matters: Any site with an on-site search bar is a potential target for this abuse, and the fix Google actually recommends, noindex on auto-generated search-result pages rather than just a robots.txt block, is a check most sites have never run.Fri, 31 Jul 2026 00:00:00 GMThttps://brandonlazovic.dev/pulse/2026-07-31/#google-ai-generated-shopping-ad-descriptionshttps://brandonlazovic.dev/pulse/2026-07-31/#google-ai-generated-shopping-ad-descriptionsPPC practitioner Brodie Clark spotted Google testing AI-generated descriptions on Shopping and Product ads in late July 2026, extending a Search-ads experiment Google confirmed earlier in the month as a small test of whether AI-generated context helps shoppers make more informed decisions. Google has not confirmed the Shopping-ads expansion or said whether it will roll out beyond the current test, and advertisers currently have no way to edit or opt out of the generated text. Why it matters: Shopping advertisers who spend heavily optimizing titles and descriptions for click-through now have a layer sitting on top of that work that they cannot directly control.Fri, 31 Jul 2026 00:00:00 GMThttps://brandonlazovic.dev/pulse/2026-07-30/#similarweb-ai-overviews-40-percenthttps://brandonlazovic.dev/pulse/2026-07-30/#similarweb-ai-overviews-40-percentSimilarweb's 2026 Generative AI Landscape Report states that more than 40% of US searches now trigger an AI Overview, up sharply from roughly 15% a year earlier. Search Engine Roundtable, covering the report on July 30, 2026, cited a more precise Similarweb figure of about 43% and a separate Semrush estimate near 48%, noting informational queries run higher than the average and that AI Mode visits are growing even faster than AI Overview appearances. Why it matters: An AI Overview showing on a plurality of queries is the SERP practitioners now optimize against by default, not the exception on a handful of head terms.Thu, 30 Jul 2026 00:00:00 GMThttps://brandonlazovic.dev/pulse/2026-07-30/#google-cloud-snowflake-iceberg-catalog-federationhttps://brandonlazovic.dev/pulse/2026-07-30/#google-cloud-snowflake-iceberg-catalog-federationAt Google Cloud Next Tokyo on July 29, 2026, Google Cloud put Catalog Federation into preview, letting its Lakehouse runtime catalog and Snowflake's Horizon Catalog read and write the same Apache Iceberg tables without copying data. Snowflake's own post confirms the integration: its Catalog-Linked Database auto-discovers Google Cloud's Iceberg tables via the open Iceberg REST Catalog standard, using vended, short-lived credentials instead of static IAM keys. AWS Glue and Databricks Unity Catalog get the same federation. Why it matters: Merchants and publishers running product or content data across Snowflake and BigQuery lose a common reason to keep duplicate copies of the same catalog just so each platform's AI tools can see it.Thu, 30 Jul 2026 00:00:00 GMThttps://brandonlazovic.dev/pulse/2026-07-30/#shopify-whatsapp-messaging-nativehttps://brandonlazovic.dev/pulse/2026-07-30/#shopify-whatsapp-messaging-nativeShopify said on July 29, 2026 that its Shopify Messaging app now supports building WhatsApp marketing campaigns natively, using pre-built or custom templates, product images pulled from the merchant's own catalog, interactive reply buttons, and keyword-triggered automated responses. Merchants pay per message sent rather than a flat subscription. Shopify positions the feature as a channel to reach customers globally outside email and SMS, without naming plan-tier or regional restrictions in the announcement. Why it matters: A native, pay-per-message WhatsApp channel removes one more reason merchants keep paying for a standalone messaging app just to run a campaign outside email and SMS.Thu, 30 Jul 2026 00:00:00 GMThttps://brandonlazovic.dev/pulse/2026-07-29/#google-search-console-platform-properties-livehttps://brandonlazovic.dev/pulse/2026-07-29/#google-search-console-platform-properties-liveGoogle confirmed on July 29, 2026 that platform properties in Search Console, first announced earlier this month, are now globally available to everyone. The feature tracks how a site's Instagram, TikTok, X, and YouTube content performs in Google Search, Discover, and Google News. Alongside the rollout, Google published a new guide on analyzing social and video performance, covering query-group trends in the Insights report, a 24-hour spike filter, cross-platform data export, and annotations for tracking edits like a retitled video. Why it matters: The property launched as a gradual beta three weeks ago, and what's new today is a defined workflow for the data: spotting trending query groups, catching 24-hour spikes, and exporting cross-platform comparisons instead of treating the report as a novelty.Wed, 29 Jul 2026 00:00:00 GMThttps://brandonlazovic.dev/pulse/2026-07-29/#gemini-api-managed-agents-3-6-flash-hookshttps://brandonlazovic.dev/pulse/2026-07-29/#gemini-api-managed-agents-3-6-flash-hooksOn July 28, 2026, Google said the antigravity-preview-05-2026 agent in its Gemini API Managed Agents now runs Gemini 3.6 Flash by default, though developers can still select Gemini 3.5 Flash or the lower-cost 3.5 Flash-Lite instead. Google also added environment hooks that run a custom script before or after a tool call inside the sandbox, a max_total_tokens budget cap, cron-based scheduled triggers, and a free tier for testing without billing. Why it matters: Hooks that inspect a tool call before it runs are the first native audit checkpoint Google has shipped for these agent sandboxes.Wed, 29 Jul 2026 00:00:00 GMThttps://brandonlazovic.dev/pulse/2026-07-29/#databricks-ai-search-target-qpshttps://brandonlazovic.dev/pulse/2026-07-29/#databricks-ai-search-target-qpsOn July 28, 2026, Databricks said standard AI Search endpoints can now scale by setting a single target_qps parameter, without manually provisioning replicas, sizing nodes, or building a load balancer. It targets retrieval-heavy paths like site search bars, 'you may also like' recommendation panels, and real-time entity-resolution lookups, where every page view can trigger a query and unmanaged capacity produces 429 errors or latency spikes. Why it matters: Teams running product-catalog or recommendation search on Databricks can skip a capacity-planning step that used to require dedicated infrastructure work before a retrieval feature could handle real traffic.Wed, 29 Jul 2026 00:00:00 GMThttps://brandonlazovic.dev/pulse/2026-07-29/#shopify-ups-return-labelshttps://brandonlazovic.dev/pulse/2026-07-29/#shopify-ups-return-labelsOn July 29, 2026, Shopify said merchants can now generate a UPS return label directly from any fulfilled order in the Shopify admin, choosing either Shopify's discounted UPS rates or their own UPS account, then deliver it to the customer by email or a shareable link. The feature covers U.S. domestic orders only. Carriers charge only once they scan a label, and unused labels expire after six months at no cost. Why it matters: Returns friction raises cart abandonment and lowers repeat-purchase rates, so folding label creation into the same admin removes a dependency many merchants currently pay a separate returns app to handle.Wed, 29 Jul 2026 00:00:00 GMThttps://brandonlazovic.dev/pulse/2026-07-29/#google-unavailable-after-recrawl-timinghttps://brandonlazovic.dev/pulse/2026-07-29/#google-unavailable-after-recrawl-timingResponding on LinkedIn to a practitioner question about dynamically extending unavailable_after dates on short-lived listings, Google's Gary Illyes said pushing the date forward works in principle, but Google must crawl the page again to register the new date. Miss that recrawl before the original date passes, he said, and the URL can drop from the index on the old schedule anyway. He suggested setting the tag to the real expiration date instead of a placeholder that keeps moving. Why it matters: Sites relying on unavailable_after for high-turnover inventory such as renewed listings, ads, or time-limited product pages risk premature deindexing if their crawl frequency lags how often the date changes.Wed, 29 Jul 2026 00:00:00 GMThttps://brandonlazovic.dev/pulse/2026-07-29/#bing-testing-pricing-product-imageshttps://brandonlazovic.dev/pulse/2026-07-29/#bing-testing-pricing-product-imagesA screenshot posted on X by Sachin Patel showed Bing overlaying a price directly on a product image inside organic search results, where Bing normally shows only a star rating on the image, with the price listed separately underneath. Search Engine Roundtable's Barry Schwartz says he separately saw a second layout, with two rows of product images stacked under a single result snippet. Bing has not confirmed either change, and Schwartz notes Bing runs a lot of SERP feature tests. Why it matters: A price shown directly on the product thumbnail changes what makes a listing image worth clicking, so anyone optimizing product images for Bing's organic results should watch this as a live test, not yet a ranking factor.Wed, 29 Jul 2026 00:00:00 GMThttps://brandonlazovic.dev/pulse/2026-07-28/#claude-chats-indexed-disallow-not-noindexhttps://brandonlazovic.dev/pulse/2026-07-28/#claude-chats-indexed-disallow-not-noindexShared Claude chat links surfaced in Google search over the weekend of July 25-26, 2026, exposing medical records, internal documents, and children's names in some chats, per TechCrunch. Search Engine Journal's July 27 check of claude.ai's robots.txt found /share/* disallowed for all crawlers, yet those same URLs return an X-Robots-Tag: none (noindex) header Google can't read because it's blocked from crawling the page. Anthropic says only publicly posted links get indexed; unsharing a chat doesn't remove it from Google's index. Why it matters: For any site using robots.txt to hide sensitive paths, this is a reminder that disallow only stops crawling, not indexing, and noindex only works if the crawler is allowed to read it.Tue, 28 Jul 2026 00:00:00 GMThttps://brandonlazovic.dev/pulse/2026-07-28/#google-ads-api-passkey-mandatoryhttps://brandonlazovic.dev/pulse/2026-07-28/#google-ads-api-passkey-mandatoryStarting August 5, 2026, Google Ads API's user authentication workflow will require a passkey to generate new OAuth 2.0 refresh tokens, replacing passwords and SMS/TOTP two-factor codes, per Google's Ads Developer Blog. Existing refresh tokens keep working unchanged. New passkeys carry a 7-day security delay before they're trusted. The requirement extends to Google Ads Editor, Ads Scripts, BigQuery Data Transfer Service, and Data Studio; service-account workflows are unaffected. Why it matters: Agencies and SaaS platforms that mint OAuth tokens on behalf of Google Ads clients need a passkey enrolled well before August 5 to avoid onboarding delays from the 7-day trust window.Tue, 28 Jul 2026 00:00:00 GMThttps://brandonlazovic.dev/pulse/2026-07-28/#snowflake-cortex-ai-gateway-mcp-governancehttps://brandonlazovic.dev/pulse/2026-07-28/#snowflake-cortex-ai-gateway-mcp-governanceAt Black Hat 2026, Snowflake announced Cortex AI Gateway, a centralized layer built by integrating Natoma's MCP gateway to enforce identity, policy, and audit at the tool-call level for AI agents, covering both Snowflake-native tools and third-party ecosystems like Amazon Bedrock, Azure AI Foundry, ChatGPT, Claude Code, and Cursor. Alongside it, Snowflake moved Agent Identity tracking and AI Security Posture Management to general availability, and launched a new Data Exfiltration Prevention package into preview. Why it matters: MCP tool-call governance is becoming a checked box on enterprise AI security reviews, and this is a major data-cloud vendor bundling it directly into the platform rather than leaving it to a point solution.Tue, 28 Jul 2026 00:00:00 GMThttps://brandonlazovic.dev/pulse/2026-07-28/#sap-bdc-connect-bigquery-gahttps://brandonlazovic.dev/pulse/2026-07-28/#sap-bdc-connect-bigquery-gaSAP and Google Cloud announced general availability of SAP Business Data Cloud (BDC) Connect for BigQuery on July 27, 2026, giving zero-copy, bidirectional access between SAP data products and BigQuery without replicating or copying data. The connector supports SAP Business Data Cloud instances on Google Cloud and AWS, with Azure-hosted support coming soon, and is aimed at grounding AI agents (Gemini Enterprise, SAP Joule) in live operational data rather than stale exports. Why it matters: Zero-copy cross-platform data access removes one more excuse for stale product or inventory data feeding AI shopping and search surfaces, if your stack touches both SAP and BigQuery.Tue, 28 Jul 2026 00:00:00 GMThttps://brandonlazovic.dev/pulse/2026-07-28/#shopify-payments-multicurrency-payout-limit-removedhttps://brandonlazovic.dev/pulse/2026-07-28/#shopify-payments-multicurrency-payout-limit-removedShopify removed the 8-currency cap on bank accounts for multi-currency payouts in Shopify Payments as of July 27, 2026; merchants can now add one bank account per supported payout currency their region and plan allow. Multi-currency payouts let merchants receive customer payments in the currency charged, avoiding a currency conversion back to their home currency and keeping balances in currencies they already use to pay suppliers, staff, and taxes abroad. Fees still apply to non-domestic-currency payouts. Why it matters: Merchants selling into many currencies can now cut FX conversion drag on every payout, not just a chosen 8, directly improving international margin without touching pricing or feeds.Tue, 28 Jul 2026 00:00:00 GMThttps://brandonlazovic.dev/pulse/2026-07-28/#shopify-pos-nearby-device-loginhttps://brandonlazovic.dev/pulse/2026-07-28/#shopify-pos-nearby-device-loginShopify POS launched quick nearby device login on July 27, 2026: a staff member with the POS Device Setup role can approve a new device from one already signed in, detected over Bluetooth (or via QR code scan without Bluetooth), then enter a store manager PIN, replacing manual credential entry on busy floors. New devices sign in for the current day only by default; merchants can enable a Forever session option in POS channel admin for longer sessions. Why it matters: Faster device onboarding at the register cuts checkout-line friction during peak traffic, a direct conversion lever for brick-and-mortar retailers running Shopify POS at scale.Tue, 28 Jul 2026 00:00:00 GMThttps://brandonlazovic.dev/pulse/2026-07-28/#shopify-pos-cart-sharinghttps://brandonlazovic.dev/pulse/2026-07-28/#shopify-pos-cart-sharingShopify released cart sharing for Shopify POS on July 27, 2026 (version 11.11, POS Pro only): any team member can pick up, continue, or close a colleague's cart from another device, with carts saving automatically and a view filtered to the signed-in staff member by default. This enables floor-to-register selling: an associate builds a cart with a customer, then hands it off for someone else to close at the register; any cart converts to a draft order in one action. Why it matters: Cart continuity across devices removes a common line-loss point: a customer who was mid-cart on the floor no longer has to restart at the register.Tue, 28 Jul 2026 00:00:00 GMThttps://brandonlazovic.dev/pulse/2026-07-28/#google-search-console-indexing-report-delayshttps://brandonlazovic.dev/pulse/2026-07-28/#google-search-console-indexing-report-delaysGoogle Search Console's page indexing report has been intermittently stuck since June 11, 2026, per Search Engine Roundtable's Barry Schwartz, who has logged the gaps daily and says it recurred again on July 27. Data collapses into weekly, not daily, snapshots across three stretches so far: June 13-30 (18 days), July 1-10 (10 days), and July 11-24 (14 days), corroborated by SEO Brodie Clark's own account data on X. Google has not commented or acknowledged the issue. Why it matters: If your indexing-status debugging depends on day-over-day GSC deltas right now, budget for the report lagging weeks behind, not days, until Google fixes the pipeline.Tue, 28 Jul 2026 00:00:00 GMThttps://brandonlazovic.dev/pulse/2026-07-27/#chatgpt-ads-manager-average-daily-budget-shifthttps://brandonlazovic.dev/pulse/2026-07-27/#chatgpt-ads-manager-average-daily-budget-shiftOpenAI's Ads Manager help center confirms that starting July 27, 2026, daily budgets become a seven-day rolling average rather than a fixed daily cap: spend can run up to double the budget on a given day, capped at seven times it per week, with no advertiser action required. OpenAI's advertiser email, quoted independently by Search Engine Roundtable and Search Engine Land, also details conversion-optimized (oCPC) bidding, geo-exclusion targeting, a Bulk API, and refreshed product-feed ads showing price and star ratings. Why it matters: Advertisers who read a daily budget as a hard per-day ceiling will see individual days run up to double that amount, and the new product-feed ad format puts feed data quality on display inside the ad unit itself.Mon, 27 Jul 2026 00:00:00 GMThttps://brandonlazovic.dev/pulse/2026-07-27/#google-request-indexing-robots-txt-bughttps://brandonlazovic.dev/pulse/2026-07-27/#google-request-indexing-robots-txt-bugA practitioner reported on Bluesky that Search Console's request-a-recrawl action inside the robots.txt report was throwing an unknown error on repeated attempts across several properties. Google's John Mueller replied that the feature worked on his own site at first, then posted directly on Bluesky that he could reproduce the same error on some of his other sites and would look into it. No official incident notice or fix has been published as of this writing. Why it matters: Anyone hitting this error on a robots.txt report shouldn't assume it's specific to their account or permissions, since a Google engineer has independently reproduced the same failure.Mon, 27 Jul 2026 00:00:00 GMThttps://brandonlazovic.dev/pulse/2026-07-25/#google-review-snippet-incentivized-reviews-banhttps://brandonlazovic.dev/pulse/2026-07-25/#google-review-snippet-incentivized-reviews-banGoogle updated its review snippet structured data guidelines on July 24, 2026 to explicitly prohibit fake or undisclosed incentivized reviews, both on the page and in the markup itself. The guidance names two disqualifying patterns: reviews not based on a genuine experience, and reviews written in exchange for money, discounts, vouchers, or free products without clear, prominent disclosure. Google can revoke rich-result eligibility for pages that violate review schema policy. Why it matters: Any site pulling Review or AggregateRating markup from an incentivized review program now has a documented, citable reason Google can strip its star ratings from search results.Sat, 25 Jul 2026 00:00:00 GMThttps://brandonlazovic.dev/pulse/2026-07-25/#shopify-app-added-analytics-annotationshttps://brandonlazovic.dev/pulse/2026-07-25/#shopify-app-added-analytics-annotationsShopify shipped a new capability on July 24, 2026 letting installed apps place annotations directly onto a merchant's analytics charts, marking events like a product launch, marketing campaign, discount, checkout offer, or supplier change. Each annotation shows the originating app's name and icon. Annotations do not alter underlying report data; they exist so merchants can visually correlate business events against metrics like sales, sessions, and conversion rate. Why it matters: This gives e-commerce teams a native way to line up a content, pricing, or catalog change with the exact moment a metric moved, instead of reconstructing timelines from separate changelogs.Sat, 25 Jul 2026 00:00:00 GMThttps://brandonlazovic.dev/pulse/2026-07-25/#anthropic-launches-claude-opus-5https://brandonlazovic.dev/pulse/2026-07-25/#anthropic-launches-claude-opus-5Anthropic released Claude Opus 5 on July 24, 2026, priced at $5 per million input tokens and $25 per million output tokens, unchanged from Opus 4.8. Anthropic says it roughly doubles Opus 4.8's Frontier-Bench score, lands within 0.5% of Fable 5's coding benchmark at half the cost, and becomes the default model on Claude Max and the strongest option on Claude Pro. A new effort toggle lets users trade intelligence for speed and cost. Why it matters: Opus 5 is the model most agentic SEO and e-commerce pipelines, retrieval, extraction, and classification tasks alike, will get benchmarked against for both quality and cost starting now.Sat, 25 Jul 2026 00:00:00 GMThttps://brandonlazovic.dev/pulse/2026-07-25/#claude-opus-5-snowflake-cortex-ai-previewhttps://brandonlazovic.dev/pulse/2026-07-25/#claude-opus-5-snowflake-cortex-ai-previewSnowflake announced public preview availability of Claude Opus 5 inside Cortex AI on July 24, 2026, the same day Anthropic released the model. Opus 5 powers Snowflake's CoCo coding agent, its CoWork personal work agents with a Deep Research mode, and Cortex AI Functions for SQL-based multimodal analysis, all running inside Snowflake's governed security perimeter with adjustable reasoning-effort levels. Why it matters: Teams already running SEO or e-commerce data pipelines inside Snowflake can now call a frontier model for classification, extraction, or agentic analysis without moving governed data outside Snowflake's security perimeter.Sat, 25 Jul 2026 00:00:00 GMThttps://brandonlazovic.dev/pulse/2026-07-25/#google-open-knowledge-format-v0-2-trust-signalshttps://brandonlazovic.dev/pulse/2026-07-25/#google-open-knowledge-format-v0-2-trust-signalsGoogle Cloud published Open Knowledge Format v0.2 on July 24, 2026, adding frontmatter fields for provenance, verification tiers (unverified, machine-confirmed, human-reviewed), absolute-date freshness, lifecycle status, and a new Attested Computation type that proves a query ran the sanctioned code rather than an agent's improvised version. The update is backward compatible: v0.1 bundles work unchanged, and reference implementations plus sample bundles shipped alongside it. Why it matters: As more of the data feeding AI answers gets written by agents rather than people, this gives data teams a standard way to mark which of that content has actually been checked before something downstream cites it as fact.Sat, 25 Jul 2026 00:00:00 GMThttps://brandonlazovic.dev/pulse/2026-07-24/#databricks-unity-ai-gateway-spend-controlshttps://brandonlazovic.dev/pulse/2026-07-24/#databricks-unity-ai-gateway-spend-controlsDatabricks announced on July 23, 2026 that Unity AI Gateway now includes AI spend controls: budget alerts and hard caps set per user, per use case, per workspace, or organization-wide (for example, $2,000 per user monthly, or $200,000 org-wide across every model). A Budgets dashboard tracks spending trends and per-user status, and Unity Catalog system tables log every request's cost by identity, workspace, model, and provider. Hard caps automatically stop requests once a budget is exceeded. Why it matters: Any team running LLM pipelines or agents against Databricks-hosted models gets a native way to stop a runaway retry loop from turning into an uncapped bill.Fri, 24 Jul 2026 00:00:00 GMThttps://brandonlazovic.dev/pulse/2026-07-24/#google-ai-max-unlocks-new-monetizable-searcheshttps://brandonlazovic.dev/pulse/2026-07-24/#google-ai-max-unlocks-new-monetizable-searchesOn Alphabet's Q2 2026 earnings call, SVP Philipp Schindler said Gemini's improved query understanding is unlocking billions of net new searches that were not really monetizable before, by matching ads to long, conversational queries that keyword targeting could not reach. He said advertisers who adopt AI Max or Performance Max see an average 15% more conversions or value on Search at a similar ROAS, and that Shopping ads saw a 20% improvement in relevance on complex queries. Why it matters: Ad inventory is expanding into query territory that used to be organic-only or too ambiguous to target at all, which raises the stakes for whether a brand's own content or an advertiser's bid wins a conversational answer.Fri, 24 Jul 2026 00:00:00 GMThttps://brandonlazovic.dev/pulse/2026-07-24/#yelp-reviews-license-to-chatgpthttps://brandonlazovic.dev/pulse/2026-07-24/#yelp-reviews-license-to-chatgptYelp signed a licensing deal giving OpenAI access to its reviews, ratings, photos, and business details so ChatGPT can answer local queries with real-time recommendation data, Yelp CEO Jeremy Stoppelman told Axios exclusively on July 23, 2026. Yelp's branding and links will appear when ChatGPT uses its content, and a Request a Quote feature is coming soon for contacting local service providers inside ChatGPT. Terms were undisclosed, and the deal does not block Yelp from licensing to other AI platforms. Why it matters: Local, service, and product recommendations are moving into a chat answer that cites Yelp directly instead of sending a click through to a Yelp listing or an SEO-optimized local page.Fri, 24 Jul 2026 00:00:00 GMThttps://brandonlazovic.dev/pulse/2026-07-24/#ai-max-spotted-standard-shopping-campaignshttps://brandonlazovic.dev/pulse/2026-07-24/#ai-max-spotted-standard-shopping-campaignsPaid-search practitioner Arpan Banerjee posted LinkedIn screenshots on July 23, 2026 showing AI Max capabilities, conversational query matching, auto-generated copy, Final URL Expansion, and the option to serve Shopping or text ads based on intent, appearing inside Standard Shopping campaigns, a type that previously kept more manual advertiser control than Performance Max. Thomas Eccel posted separately, reported the next day, adding an unspecified conversion-lift estimate from combining asset optimization features. Google has not confirmed the rollout or published documentation. Why it matters: Standard Shopping campaigns were the fallback for advertisers who wanted to avoid Performance Max's automation and opacity, and AI Max folding into that campaign type narrows where manual control is still available.Fri, 24 Jul 2026 00:00:00 GMThttps://brandonlazovic.dev/pulse/2026-07-24/#health-in-chatgpt-launcheshttps://brandonlazovic.dev/pulse/2026-07-24/#health-in-chatgpt-launchesOpenAI announced on July 23, 2026 that Health in ChatGPT is rolling out to logged-in US users age 18 and up on Free, Go, Plus, and Pro plans across web and iOS. Users can connect Apple Health and supported medical records from US hospital systems, One Medical, or Function Health so ChatGPT can reference labs, medications, and activity in any conversation, not just a separate Health tab. Connected health data is never used to train models or target ads. Why it matters: OpenAI just gave itself a persistent, permissioned data layer competitors don't have, and that is a moat health and wellness brands will have to design content and product data around.Fri, 24 Jul 2026 00:00:00 GMThttps://brandonlazovic.dev/pulse/2026-07-23/#gemini-live-camera-assistanthttps://brandonlazovic.dev/pulse/2026-07-23/#gemini-live-camera-assistantGoogle announced on July 22, 2026 that Gemini Live now answers questions about anything a phone's camera sees or a shared screen displays. Users tap the Live icon, then the camera or screen-share control, to get step-by-step help with tasks like appliance repairs, organizing a space, navigating an unfamiliar app, comparing items while shopping, or brainstorming from a visual reference, without typing a query first. Why it matters: Camera-first assistance moves purchase-decision moments into a channel product pages were never built to serve, since the model reads the physical object or screen directly instead of a search result.Thu, 23 Jul 2026 00:00:00 GMThttps://brandonlazovic.dev/pulse/2026-07-23/#google-ads-video-campaign-groups-globalhttps://brandonlazovic.dev/pulse/2026-07-23/#google-ads-video-campaign-groups-globalGoogle made video campaign groups globally available in Google Ads on July 13, 2026, letting advertisers coordinate reach and frequency across multiple YouTube campaigns under one group while keeping campaign-level budgets and creative separate. The group view adds unified reporting on unique reach and average weekly impressions. Google cited its own Meridian research showing a 2.7-per-week optimal frequency drove a 19% ROI lift, and said the feature is coming to Display & Video 360 next. Why it matters: Reach and frequency now get measured and capped at the group level instead of per campaign, changing how e-commerce advertisers benchmark YouTube ad exposure and overexposure risk.Thu, 23 Jul 2026 00:00:00 GMThttps://brandonlazovic.dev/pulse/2026-07-23/#google-ads-api-v25https://brandonlazovic.dev/pulse/2026-07-23/#google-ads-api-v25Google released Google Ads API v25 on July 22, 2026. New capabilities include YouTube non-skippable ad breakdowns by duration (standard, max 30s, max 60s), YouTube Shorts social metrics (comments, likes, shares) across Ad Group, Asset, Campaign, and Video resources, a new loyalty retention goal for program-member retention, a revamped customer acquisition goal replacing the legacy CustomerLifecycleGoal resource, and consent-gated YouTube channel insights for creators sharing non-public audience data. Developers must upgrade client libraries to use the new fields. Why it matters: Every agency's automated bidding and attribution tooling built on the API needs a library upgrade before the new YouTube and loyalty segments become usable, which sets the pace for when reporting can actually reflect them.Thu, 23 Jul 2026 00:00:00 GMThttps://brandonlazovic.dev/pulse/2026-07-23/#pypi-restricts-file-uploads-after-14-dayshttps://brandonlazovic.dev/pulse/2026-07-23/#pypi-restricts-file-uploads-after-14-daysPyPI announced on July 22, 2026 that it now rejects new files added to a release once that release passes 14 days old. The change closes a theoretical supply-chain gap: if a maintainer's publishing credentials were ever compromised, an attacker could previously have injected a malicious file into an already-trusted, widely-depended-on release without triggering a new-version alert. PyPI's Seth Larson said there is no evidence the gap was ever exploited. Why it matters: Any pipeline that relies on adding a late file to an old release, rare but real for delayed wheel builds, now fails and needs its packaging schedule fixed before that release passes the 14-day window.Thu, 23 Jul 2026 00:00:00 GMThttps://brandonlazovic.dev/pulse/2026-07-23/#chatgpt-ads-manager-conversion-tooltip-spottedhttps://brandonlazovic.dev/pulse/2026-07-23/#chatgpt-ads-manager-conversion-tooltip-spottedPaid-media practitioner Craig Graham reported on LinkedIn, via Search Engine Roundtable on July 23, 2026, that hovering over the Conversions column in ChatGPT Ads Manager now surfaces a tooltip breaking the value into View Through Attribution (VTA) and Click Through Attribution (CTA) segments. OpenAI's own Ads Manager help documentation describes segmenting conversion columns by event type but had not been updated to document this specific VTA/CTA tooltip as of this writing. Why it matters: Attribution-model transparency in a fast-growing, thinly-documented ad platform determines whether advertisers can trust reported ROI enough to shift budget into ChatGPT ads at the same scale as Google or Meta.Thu, 23 Jul 2026 00:00:00 GMThttps://brandonlazovic.dev/pulse/2026-07-22/#gemini-3-6-flash-family-launchhttps://brandonlazovic.dev/pulse/2026-07-22/#gemini-3-6-flash-family-launchGoogle released three Gemini models on July 21, 2026. Gemini 3.6 Flash, priced at $1.50/$7.50 per million input/output tokens, cuts output token use 17% and gains ground on coding and computer-use benchmarks. Gemini 3.5 Flash-Lite runs 350 tokens per second at $0.30/$2.50 per million tokens and is rolling into Google Search. Gemini 3.5 Flash Cyber is a cybersecurity-specific model restricted to governments and trusted partners through CodeMender. Why it matters: Cheaper, faster Flash-Lite reaching Google Search directly shapes the model doing the reasoning behind AI Overviews and AI Mode answers.Wed, 22 Jul 2026 00:00:00 GMThttps://brandonlazovic.dev/pulse/2026-07-22/#google-crawl-budget-doc-conservative-defaulthttps://brandonlazovic.dev/pulse/2026-07-22/#google-crawl-budget-doc-conservative-defaultGoogle's Search Central documentation on crawl budget was updated on July 22, 2026, to state explicitly that every site begins with an identical, conservative crawl capacity limit. If there is demand to crawl more and the site stays healthy on server response consistency and latency, Google's systems automatically raise that limit over time. The capacity limit is shared across all of Google's crawlers, so high demand from one crawler product reduces the room available to others. Why it matters: A shared, demand-gated crawl ceiling means server health and content demand, not raw site size, decide how much of a site Google actually indexes.Wed, 22 Jul 2026 00:00:00 GMThttps://brandonlazovic.dev/pulse/2026-07-22/#google-serpapi-dmca-dismissedhttps://brandonlazovic.dev/pulse/2026-07-22/#google-serpapi-dmca-dismissedOn July 20, 2026, Chief Judge Yvonne Gonzalez Rogers of the U.S. District Court for the Northern District of California granted SerpApi's motion to dismiss Google's DMCA anti-circumvention claims. The dismissal is with prejudice where Google's SearchGuard system controlled access to results containing no copyrighted content, but without prejudice where a copyrighted component was involved, giving Google 21 days to refile a narrower claim. SerpApi confirmed the ruling on its own blog and published the court's order. Why it matters: A DMCA anti-circumvention theory failing against a scraping tool narrows one of the few legal levers platforms have used to restrict automated access to search results.Wed, 22 Jul 2026 00:00:00 GMThttps://brandonlazovic.dev/pulse/2026-07-22/#google-ads-missed-growth-recommendations-tabhttps://brandonlazovic.dev/pulse/2026-07-22/#google-ads-missed-growth-recommendations-tabSearch Engine Land reports Google Ads is rolling out a beta to eligible advertisers that surfaces Missed Opportunity Reporting inside the Recommendations tab, showing modeled estimates of clicks, conversions, and conversion value lost to budget or bid constraints. Google's own product page confirms the underlying capability, interactive weekly charts comparing actual versus potential performance across Search, Shopping, Performance Max, Demand Gen, YouTube, App, and Hotel campaigns, but does not itself confirm tab placement or beta scope. Why it matters: Surfacing modeled missed revenue right where advertisers already act on recommendations would push budget increases through Google's own framing rather than independent analysis.Wed, 22 Jul 2026 00:00:00 GMThttps://brandonlazovic.dev/pulse/2026-07-22/#openai-presence-enterprise-agentshttps://brandonlazovic.dev/pulse/2026-07-22/#openai-presence-enterprise-agentsOpenAI introduced Presence on July 22, 2026, a product for deploying AI agents that resolve customer and internal workflows such as billing issues, insurance claims, and IT requests. Each deployment is scoped to a single job with only the access that job requires. The company sets approval and escalation policies, and OpenAI's Codex proposes updates to the agent as production sessions reveal gaps. Presence supports real-time voice and chat today. Why it matters: A packaged, governed agent-deployment product signals OpenAI is standardizing enterprise agent rollout rather than leaving it to each company's custom build.Wed, 22 Jul 2026 00:00:00 GMThttps://brandonlazovic.dev/pulse/2026-07-22/#chatgpt-small-business-programhttps://brandonlazovic.dev/pulse/2026-07-22/#chatgpt-small-business-programOpenAI announced the ChatGPT for small business program on July 21, 2026. It combines hands-on virtual training webinars, in-person AI academies run with local business owners, new getting-started guides, and curated partner integrations from Dropbox, Shopify, Intuit, Slack, Atlassian, and Wix. OpenAI cited last year's Small Business AI Jams, where 78% of attendees built a working AI workflow in a day and 42% saved more than five hours weekly. Why it matters: Direct Shopify and Intuit integration paths give small e-commerce merchants a guided on-ramp to agent workflows, an audience that otherwise would not build this tooling itself.Wed, 22 Jul 2026 00:00:00 GMThttps://brandonlazovic.dev/pulse/2026-07-22/#google-ads-ga4-bulk-linkinghttps://brandonlazovic.dev/pulse/2026-07-22/#google-ads-ga4-bulk-linkingA bulk-linking option has appeared in Google Ads under Data Manager, Google Analytics 4, Link Setup, letting advertisers select or deselect multiple Google Ads accounts to connect to one GA4 property in a single action instead of linking each account individually. Paid-search practitioner Thomas Eccel spotted the change and Search Engine Land reported it. Google's own GA4 linking help documentation had not been updated to reflect it as of this writing. Why it matters: Agencies and franchises managing many ad accounts against one GA4 property would lose a recurring manual-linking chore, if the observed behavior is a permanent rollout rather than a test.Wed, 22 Jul 2026 00:00:00 GMThttps://brandonlazovic.dev/pulse/2026-07-21/#google-ads-lead-form-eligibility-loweredhttps://brandonlazovic.dev/pulse/2026-07-21/#google-ads-lead-form-eligibility-loweredGoogle's Lead Form assets help documentation has dropped the $50,000 lifetime-ad-spend path that previously let advertisers qualify without meeting its reputation-and-verification standard, per Search Engine Land's July 20, 2026 review of the current documentation. The remaining path requires a policy-compliant reputation plus more than $1,000 spent in an account, or $15,000 across accounts. The same documentation still lists Display campaigns under requirements even though its overview drops Display as a supported campaign type, an inconsistency Google has not resolved. Why it matters: Advertisers who used to qualify for Lead Forms purely by hitting $50,000 in lifetime spend now need to clear Google's reputation-and-verification bar instead, and the conflicting Display-campaign guidance means anyone running lead forms on Display should confirm eligibility directly in-account rather than trust the docs.Tue, 21 Jul 2026 00:00:00 GMThttps://brandonlazovic.dev/pulse/2026-07-21/#google-ads-enhanced-brand-lift-studieshttps://brandonlazovic.dev/pulse/2026-07-21/#google-ads-enhanced-brand-lift-studiesAdvertisers can now choose between Standard and a new Enhanced option for Brand Lift Studies in Google Ads, per a LinkedIn post by Thomas Eccel that Search Engine Land reported on July 20, 2026. Enhanced Brand Lift detects campaign lift as low as 1.2 percent, versus Standard's 2 percent threshold, raising the odds of detecting a real positive lift by 60 percent. The tradeoff: Enhanced studies need roughly three times the spend of a Standard study. Why it matters: A lower detection floor only pays off for advertisers with the budget to hit Enhanced's roughly 3x spend requirement, so this mostly benefits large brand campaigns rather than mid-market accounts.Tue, 21 Jul 2026 00:00:00 GMThttps://brandonlazovic.dev/pulse/2026-07-21/#aws-sagemaker-unified-studio-redshift-cross-accounthttps://brandonlazovic.dev/pulse/2026-07-21/#aws-sagemaker-unified-studio-redshift-cross-accountAWS published a data-mesh pattern on July 20, 2026, that lets Amazon SageMaker Unified Studio govern Redshift data sharing across a producer account, a consumer account, and a central governance account. Producers publish Redshift tables and views to a shared catalog; consumers request access, and producers approve it inside the platform instead of coordinating manually. Redshift clusters must run RA3 nodes or Redshift Serverless, and every access grant is logged through CloudTrail, Redshift audit logs, and DataZone. Why it matters: Teams running multi-account data warehouses now have an AWS-native pattern for auditable cross-account access, instead of ad-hoc IAM roles and manual approval emails.Tue, 21 Jul 2026 00:00:00 GMThttps://brandonlazovic.dev/pulse/2026-07-21/#snowflake-ai-classify-public-preview-document-intelligencehttps://brandonlazovic.dev/pulse/2026-07-21/#snowflake-ai-classify-public-preview-document-intelligenceSnowflake published a July 20, 2026 blog post detailing its Cortex AI Functions document-intelligence stack, confirming AI_CLASSIFY has moved to public preview. AI_CLASSIFY routes different document types, invoices, contracts, claims forms, to different extraction pipelines automatically, sitting ahead of AI_PARSE_DOCUMENT, AI_EXTRACT, AI_COMPLETE, and AI_EMBED in the workflow. Snowflake also shipped a prebuilt Cortex Code skill, ai-functions-pipeline-builder, that chains those four functions into a single declarative pipeline against hundreds of thousands of documents a day. Why it matters: AI_CLASSIFY's move to public preview is the piece that makes a document-RAG pipeline production-ready rather than a demo: mixed-format document sets stop needing a human to sort them before extraction can even start.Tue, 21 Jul 2026 00:00:00 GMThttps://brandonlazovic.dev/pulse/2026-07-21/#huggingface-grabette-robot-manipulation-datahttps://brandonlazovic.dev/pulse/2026-07-21/#huggingface-grabette-robot-manipulation-dataHugging Face released Grabette on July 21, 2026, an open-source handheld gripper built from a Raspberry Pi, standard cameras, and a depth sensor, about 490 euros in parts, that records human manipulation demonstrations for robot learning. A companion robotic gripper, Gripette, about 120 euros, replays the learned movements on real robots. Recordings run through a browser-based Hugging Face Spaces pipeline that converts them into LeRobot-compatible datasets; the team trained a working grasping policy from 200 demonstrations on a 7-DoF arm. Why it matters: For teams building or evaluating physical-AI data pipelines, this shows how cheap open hardware for collecting real-world training data is becoming, outside big-lab robotics budgets.Tue, 21 Jul 2026 00:00:00 GMThttps://brandonlazovic.dev/pulse/2026-07-21/#nvidia-cosmos-3-edge-world-modelhttps://brandonlazovic.dev/pulse/2026-07-21/#nvidia-cosmos-3-edge-world-modelNVIDIA published Cosmos 3 Edge on Hugging Face on July 20, 2026, a 4-billion-parameter world model for physical AI that pairs an autoregressive reasoning tower with a diffusion generation tower sharing multimodal attention layers. The model processes 640x360 observations at robot-control resolution, generates 32 actions per inference, and hits real-time control at 15 Hz on NVIDIA's Jetson Thor edge hardware. It ranks first among 4-billion-parameter models on VANTAGE-Bench and ships with a DROID policy variant post-trained on robot-manipulation tasks. Why it matters: Running a capable world model directly on edge hardware, instead of round-tripping to a cloud API, is the difference between a robot or vision agent that reacts in real time and one that lags behind its environment.Tue, 21 Jul 2026 00:00:00 GMThttps://brandonlazovic.dev/pulse/2026-07-20/#google-local-services-ads-into-google-adshttps://brandonlazovic.dev/pulse/2026-07-20/#google-local-services-ads-into-google-adsGoogle is migrating Local Services Ads into Google Ads under Performance Max. Manual cost-per-lead bidding is eliminated in favor of automatic bidding, weekly budgets convert to daily averages (divided by 7), industry-level Target CPA becomes one campaign-level target, BBB callouts are discontinued in favor of at least six other structured callouts, and the lead inbox moves into Google Ads' Conversions section. Rollout starts August 2026 with home and storefront services, expands late 2026, and reaches non-U.S. accounts in 2027. Why it matters: Local-service advertisers lose manual bid control and their dedicated dashboard on a fixed timeline, forcing a workflow change mid-campaign rather than at a moment they choose.Mon, 20 Jul 2026 00:00:00 GMThttps://brandonlazovic.dev/pulse/2026-07-20/#google-youtube-shopify-app-no-reinstall-neededhttps://brandonlazovic.dev/pulse/2026-07-20/#google-youtube-shopify-app-no-reinstall-neededA rumor had circulated that Shopify merchants needed to reinstall the Google & YouTube app before the Merchant API replaces the Content API on August 18, 2026. Google updated its Merchant Center help document's FAQ to say the opposite: no action is needed, the app should stay installed as-is, and Google handles the Content API to Merchant API migration itself. Uninstalling and reinstalling can currently cause issues that Google says it is working to fix with Shopify. Why it matters: An unnecessary uninstall/reinstall driven by a rumor could break a merchant's live product-feed connection instead of fixing anything ahead of the sunset.Mon, 20 Jul 2026 00:00:00 GMThttps://brandonlazovic.dev/pulse/2026-07-20/#content-api-extended-access-form-livehttps://brandonlazovic.dev/pulse/2026-07-20/#content-api-extended-access-form-liveGoogle posted a notice atop its Merchant API documentation confirming the Content API for Shopping sunsets August 18, 2026, and linked a request form for merchants who need additional migration time. The form asks for email, partner or company name, website, Google Cloud project IDs, and a business justification, and lets requesters choose October 15 or December 31, 2026 as an extended deadline. Why it matters: Merchants who can't finish migrating by August 18 now have a documented, official process to request more time instead of guessing at Google's tolerance for lingering Content API calls.Mon, 20 Jul 2026 00:00:00 GMThttps://brandonlazovic.dev/pulse/2026-07-20/#google-ads-ai-label-corner-placement-guidancehttps://brandonlazovic.dev/pulse/2026-07-20/#google-ads-ai-label-corner-placement-guidanceGoogle updated the help document it created July 9 when it announced labeling for ads created or edited with AI. The guidance previously told advertisers to keep labels outside a 5.5% margin around an image's full perimeter; it now says to avoid placing labels in the very corners instead, while acknowledging that across the range of rendering formats, Google cannot fully guarantee a label won't be trimmed. Why it matters: Advertisers who built label placement to the old 5.5%-margin spec should recheck it, since a rule that admits it can't guarantee trim-safety leaves real room to get technically compliant and still lose the label.Mon, 20 Jul 2026 00:00:00 GMThttps://brandonlazovic.dev/pulse/2026-07-19/#google-bot-challenge-screens-drop-canonicalhttps://brandonlazovic.dev/pulse/2026-07-19/#google-bot-challenge-screens-drop-canonicalOn the Search Off the Record podcast's July 16, 2026 episode on the Indexing Report, Google's John Mueller and Martin Splitt explained that when a CDN or security layer serves an 'are you a bot' challenge screen to Googlebot with a 200 status code, Google indexes that screen instead of the real page. Because the same generic screen appears across many unrelated sites, Google's deduplication picks one as canonical and treats the rest, including the real content, as duplicates. Why it matters: Any site running aggressive bot-detection or CDN security rules risks losing indexed pages to a competitor's identical challenge screen, and a normal browser check will never reveal it because real visitors never see the flagged version.Sun, 19 Jul 2026 00:00:00 GMThttps://brandonlazovic.dev/pulse/2026-07-19/#search-console-validate-fix-when-it-helpshttps://brandonlazovic.dev/pulse/2026-07-19/#search-console-validate-fix-when-it-helpsOn the same Search Off the Record episode, Mueller said clicking Validate Fix in a Search Console indexing report samples a handful of the flagged pages first; only if those check out clean does Google trigger a faster recrawl of the rest. He recommends it for issues hitting many pages at once, like a server or CDN misconfiguration, and says a single fixed URL should go through URL Inspection's Request Indexing instead, since batch sampling offers no benefit there. Why it matters: Teams sitting on a backlog of 'fixed' issues in Search Console get a concrete reason to batch systemic fixes before validating, rather than burning a validation cycle on every individual page.Sun, 19 Jul 2026 00:00:00 GMThttps://brandonlazovic.dev/pulse/2026-07-18/#google-ai-search-billions-clicks-no-methodologyhttps://brandonlazovic.dev/pulse/2026-07-18/#google-ai-search-billions-clicks-no-methodologyGoogle SVP Nick Fox said AI features in Search now send billions of clicks to websites weekly, on top of the billions Search sends daily overall. Neither figure comes with a baseline, denominator, or published methodology, and no earlier version of the weekly AI-specific figure appears in Google's own posts before July 17, 2026. Individual sites see only impressions in Search Console, not the clicks Google cites, so the aggregate can't be checked against any site's own data. Why it matters: A traffic claim practitioners can't reproduce from their own Search Console data isn't a metric they can act on, and the number that would actually change strategy, per-site AI referral clicks, still isn't published.Sat, 18 Jul 2026 00:00:00 GMThttps://brandonlazovic.dev/pulse/2026-07-18/#google-ai-overviews-top-stories-carouselhttps://brandonlazovic.dev/pulse/2026-07-18/#google-ai-overviews-top-stories-carouselSearch Engine Land reports a Top Stories carousel is now live inside AI Overviews for US mobile users on developing-news queries, surfacing timely articles and prioritizing a searcher's Preferred Sources when a trending story matches an outlet they've selected. The mechanism traces to a Google post from May 27, 2026 describing a 'prominent carousel' tied to Preferred Sources; that post uses 'Top Stories' only as an analogy, not this carousel's name, and doesn't confirm the scope Search Engine Land describes. Why it matters: A news carousel inside AI Overviews is a fresh visibility surface for timely content, and Preferred Sources status inside it rewards publishers a reader has already chosen over ones simply ranking well generically.Sat, 18 Jul 2026 00:00:00 GMThttps://brandonlazovic.dev/pulse/2026-07-18/#apple-maps-ads-category-restrictionshttps://brandonlazovic.dev/pulse/2026-07-18/#apple-maps-ads-category-restrictionsApple's advertising policy for its upcoming Apple Maps ads product, effective July 14, 2026, prohibits ad content promoting home services such as plumbing, electrical, locksmith, HVAC, pest control, roofing, or general contracting, along with bail bond services and cryptocurrency ATMs. Medical services ads get case-by-case review instead of an outright ban. Search Engine Journal reports a US and Canada launch this summer; Apple has not published a specific date. Why it matters: Advertisers in home services, a vertical Google's Local Services Ads was built specifically to serve, need a different paid strategy on Apple Maps once it launches, since this policy closes the category off entirely rather than gating it behind verification the way Google does.Sat, 18 Jul 2026 00:00:00 GMThttps://brandonlazovic.dev/pulse/2026-07-18/#shopify-collections-multi-source-variantshttps://brandonlazovic.dev/pulse/2026-07-18/#shopify-collections-multi-source-variantsShopify's Collections feature now supports combining automated rules, hand-picked products, exclusions, other collections, and app-sourced items within a single collection, replacing the prior either/or choice between automated and manual sourcing. Merchants can also build collections around specific variants, like a single size or color, that carry through to collection pages, storefront filters, and channels like Online Store and POS. Existing collections carry over automatically, though third-party or custom apps need API version 2026-07 to use the new capabilities. Why it matters: Variant-level and multi-source collections give merchants a native way to build precise merchandising segments, like in-stock items in one color, without the custom app logic that used to require.Sat, 18 Jul 2026 00:00:00 GMThttps://brandonlazovic.dev/pulse/2026-07-18/#bigquery-iam-data-governance-tagshttps://brandonlazovic.dev/pulse/2026-07-18/#bigquery-iam-data-governance-tagsGoogle Cloud introduced IAM data governance tags in preview, a column-level security and classification system built on Identity and Access Management that replaces BigQuery's older, region-locked policy tags. Tags are defined once org-wide, in hierarchies up to five levels deep like PII, Financial, and CreditCardNumber, then attached to columns via schema JSON or SQL and enforced through masking or raw-access policies. Because the tags are global while enforcement stays regional, they also replicate automatically to secondary regions for disaster recovery. Why it matters: Teams piping first-party or customer data into BigQuery for AI-visibility or personalization pipelines get a native way to classify and restrict sensitive columns without hand-rolled access scripts.Sat, 18 Jul 2026 00:00:00 GMThttps://brandonlazovic.dev/pulse/2026-07-18/#databricks-spark-muse-unity-ai-gatewayhttps://brandonlazovic.dev/pulse/2026-07-18/#databricks-spark-muse-unity-ai-gatewayDatabricks made Meta's Spark Muse 1.1 model available immediately through its new Model Provider Services, reachable via an OpenAI-compatible API and governed centrally through Unity AI Gateway. Unity Catalog permissions, rather than per-team API keys, now control access, while usage tracking captures token counts, latency, and cost attribution per team, and credentials for external providers stay inside Unity Catalog instead of scattered across systems. The capability is in preview across AWS, Azure, and GCP. Why it matters: Centralized model governance matters for any team running LLM-driven content, classification, or agent pipelines against Databricks-hosted data, since it replaces per-team key sprawl with one place to enforce guardrails and see real cost per workload.Sat, 18 Jul 2026 00:00:00 GMThttps://brandonlazovic.dev/pulse/2026-07-18/#nvidia-nemo-automodel-diffusers-finetuninghttps://brandonlazovic.dev/pulse/2026-07-18/#nvidia-nemo-automodel-diffusers-finetuningNVIDIA integrated its NeMo Automodel training library with Hugging Face's Diffusers, letting teams run full fine-tuning or parameter-efficient LoRA training on Hub image and video models, including FLUX.1-dev, FLUX.2-dev, Qwen-Image, Wan 2.1, and HunyuanVideo, without converting checkpoints or writing custom training code. Training scales from a single GPU to multi-node clusters using sharding like FSDP2 and tensor parallelism. NVIDIA reports roughly 35.5 images per second fine-tuning FLUX.1-dev on eight H100 GPUs, rising to about 53.7 with LoRA. Why it matters: Teams generating on-brand product imagery or video at scale now have a production path to fine-tune open diffusion models on their own catalog, instead of prompting a general-purpose model and hoping the brand look holds.Sat, 18 Jul 2026 00:00:00 GMThttps://brandonlazovic.dev/pulse/2026-07-17/#google-gemini-notebook-crawler-renamehttps://brandonlazovic.dev/pulse/2026-07-17/#google-gemini-notebook-crawler-renameNotebookLM is now Gemini Notebook, adding native code execution, a secure cloud computer per notebook, and planned integration into AI Mode in Search, while keeping its 30 million users and 600,000-plus organizations synced across Google's apps. Google's crawling documentation confirms the matching technical change: the fetcher user agent moved from Google-NotebookLM to Google-GeminiNotebook, with the old string still honored through August 2026. Why it matters: Any robots.txt allowlist, bot-management rule, or log script that hardcodes Google-NotebookLM needs the Google-GeminiNotebook string before Google retires the legacy value in August 2026.Fri, 17 Jul 2026 00:00:00 GMThttps://brandonlazovic.dev/pulse/2026-07-17/#shopify-managed-markets-eu-cancellation-returnshttps://brandonlazovic.dev/pulse/2026-07-17/#shopify-managed-markets-eu-cancellation-returnsShopify Managed Markets now supports buyer-initiated cancellation and return requests on EU-bound orders, backed by a managed 14-day cancellation and return window that satisfies the EU's Right of Withdrawal rules. Global-e, Shopify's merchant-of-record partner for Managed Markets, administers the window. Buyers request cancellations or returns through Shopify's buyer interfaces, and merchants must supply return labels but take no setup action to keep access. Why it matters: Merchants selling into the EU through Managed Markets get automatic compliance with a legal requirement, the 14-day withdrawal right, without building return logic themselves.Fri, 17 Jul 2026 00:00:00 GMThttps://brandonlazovic.dev/pulse/2026-07-17/#google-local-inventory-ads-default-behaviorhttps://brandonlazovic.dev/pulse/2026-07-17/#google-local-inventory-ads-default-behaviorSearch Engine Land reports Google will make Local Inventory Ads the default for Shopping campaigns tied to a Merchant Center account with the add-on enabled, starting August 31, 2026. The change retires the 'Local products' toggle under Other settings in favor of an Inventory filter set to Channel = Local or Online. The report traces to a Google Ads notification email surfaced by PPC specialist Arpan Banerjee; Google has not published its own announcement. Why it matters: Advertisers who rely on the 'Local products' toggle to keep local and online budgets separate need to migrate to the Inventory filter before Google flips the default, or local inventory ads could start pulling from an unintended budget.Fri, 17 Jul 2026 00:00:00 GMThttps://brandonlazovic.dev/pulse/2026-07-17/#google-storebot-help-doc-updatedhttps://brandonlazovic.dev/pulse/2026-07-17/#google-storebot-help-doc-updatedGoogle updated its help document 'How to fix: Google StoreBot crawler can't access your in-store product page' with clearer guidance on user agents, robots.txt checks, IP blocking, page speed issues, and reprocessing. The doc lists common ways merchants unintentionally block StoreBot, Googlebot, and Googlebot-image, including robots.txt rules, user-agent switching, IP blocks, firewalls, fingerprinting, and slow page loads. It also confirms that once access is restored, products typically reappear in local inventory ads within 12 to 48 hours. Why it matters: This is the doc merchants hit when local inventory ads silently stop showing; the new detail on bot-detection blocklists and fingerprinting names failure modes that generic robots.txt checks miss.Fri, 17 Jul 2026 00:00:00 GMThttps://brandonlazovic.dev/pulse/2026-07-16/#eu-google-search-data-android-interophttps://brandonlazovic.dev/pulse/2026-07-16/#eu-google-search-data-android-interopThe European Commission issued legally binding specification decisions against Google on July 15, 2026, under the Digital Markets Act. Google must share the same anonymized search data it uses to improve its own service with competing search engines, including AI chatbots with search features, under a set pricing formula. Google must also give third-party AI assistants access to core Android functions comparable to Gemini's. Why it matters: Mandated data sharing with rival search engines and AI chatbots could reshape competitive SOV analysis and how AI-visibility strategy accounts for non-Google search surfaces in the EU.Thu, 16 Jul 2026 00:00:00 GMThttps://brandonlazovic.dev/pulse/2026-07-16/#google-ai-mode-connected-appshttps://brandonlazovic.dev/pulse/2026-07-16/#google-ai-mode-connected-appsGoogle began rolling out connected apps in AI Mode the week of July 16, 2026, letting US users securely link Instacart, Canva, and YouTube Music directly inside Search. Linked apps let AI Mode add ingredients to an Instacart cart, pull design templates from Canva, and build YouTube Music playlists without leaving the results page, with more partner apps promised soon. Why it matters: Search is becoming a checkout and task-completion surface, not just an answer surface, which raises the stakes for product feed and app integration readiness.Thu, 16 Jul 2026 00:00:00 GMThttps://brandonlazovic.dev/pulse/2026-07-16/#ai-mode-personal-intelligence-calendarhttps://brandonlazovic.dev/pulse/2026-07-16/#ai-mode-personal-intelligence-calendarRobby Stein, Google Search's VP of Product, announced on July 15, 2026 that Personal Intelligence in AI Mode now connects to Google Calendar. Users can add invites or meetings to their Calendar directly from AI Mode, and responses become more tailored by accounting for a user's existing schedule. The connection is live now in the US, with more countries to follow, joining earlier Gmail and Photos links. Why it matters: Calendar awareness pushes AI Mode responses toward deeper personalization, another sign that generic content competes less well against context-aware assistant answers.Thu, 16 Jul 2026 00:00:00 GMThttps://brandonlazovic.dev/pulse/2026-07-16/#pmax-product-reporting-all-networkshttps://brandonlazovic.dev/pulse/2026-07-16/#pmax-product-reporting-all-networksGoogle Ads Help confirms that starting June 2026, Performance Max product reporting expanded beyond Search and Standard Shopping campaigns to cover all networks, including Video, App, and Demand Gen. Advertisers now get a single, comprehensive set of product-level metrics, like cost and conversions, across every eligible campaign type that pulls from Merchant Center. Google warns the change can cause a one-time jump in reported impressions and clicks. Why it matters: Unified cross-network product reporting changes what channel performance means for ecommerce advertisers reconciling Shopping data against Merchant Center feeds.Thu, 16 Jul 2026 00:00:00 GMThttps://brandonlazovic.dev/pulse/2026-07-16/#shopping-ads-free-listings-policy-mergehttps://brandonlazovic.dev/pulse/2026-07-16/#shopping-ads-free-listings-policy-mergeGoogle's Merchant Center announcements changelog states that in September 2026, it will consolidate the separate Shopping ads and free listings policies into a single set of Shopping policies for improved organization and clarity. Some individual policies will continue to apply only to Shopping ads, and Google says it will clearly flag those cases. Google states the change does not introduce more restrictive enforcement or substantive policy changes. Why it matters: A merged policy document is a paperwork change for now, but it is worth tracking in September for any quiet scope shifts between paid Shopping ads and free listings.Thu, 16 Jul 2026 00:00:00 GMThttps://brandonlazovic.dev/pulse/2026-07-16/#spark-4-2https://brandonlazovic.dev/pulse/2026-07-16/#spark-4-2Databricks announced Apache Spark 4.2 on July 16, 2026, the next release of the dominant open-source data processing engine, now available in Databricks Runtime 19 Beta. The release adds a native semantic layer for defining business metrics once, vector similarity search and geospatial types, and improved Spark Connect embedding, spanning more than 1,900 commits from 260-plus contributors. Why it matters: Engine-level releases decide what data tooling AI pipelines can rely on before the managed platforms catch up.Thu, 16 Jul 2026 00:00:00 GMThttps://brandonlazovic.dev/pulse/2026-07-16/#snowflake-iceberg-resthttps://brandonlazovic.dev/pulse/2026-07-16/#snowflake-iceberg-restSnowflake detailed how its Horizon Catalog implements the Iceberg REST Catalog standard in both directions, letting external engines read and write Snowflake-managed tables using vended, short-lived credentials instead of long-lived IAM keys. Snowflake contrasts this with Databricks Unity Catalog, which it says supports only inbound, read-only access to externally managed Iceberg tables. Why it matters: Bidirectional catalog interop reduces lock-in decisions to configuration, which changes how search-data warehouses get architected.Thu, 16 Jul 2026 00:00:00 GMThttps://brandonlazovic.dev/pulse/2026-07-15/#google-shopping-ads-14-country-expansionhttps://brandonlazovic.dev/pulse/2026-07-15/#google-shopping-ads-14-country-expansionGoogle's Merchant Center changelog confirms Shopping ads and free listings now cover 14 additional countries starting in July: Bulgaria, Bosnia and Herzegovina, Croatia, Cyprus, Estonia, Latvia, Luxembourg, Lithuania, Liechtenstein, Moldova, Montenegro, North Macedonia, Malta, and Serbia. Google frames this as new country eligibility for both ad formats, letting merchants target these markets for the first time and reach new audiences across the regions. Why it matters: Merchants selling into Southeastern Europe and the Baltics now have a legitimate new distribution channel to test before committing full catalog and feed investment there.Wed, 15 Jul 2026 00:00:00 GMThttps://brandonlazovic.dev/pulse/2026-07-15/#google-prediction-markets-ad-ban-mi-nyhttps://brandonlazovic.dev/pulse/2026-07-15/#google-prediction-markets-ad-ban-mi-nyGoogle's Prediction Markets advertising policy currently excludes Michigan, Nevada, New York, and Ohio from its approved locations list, meaning advertisers cannot run ads for exchange-listed event contracts and related products in those four states. Even in approved US locations, advertisers must hold certification as a CFTC-authorized Designated Contract Market or an NFA-authorized brokerage before Google will approve the ads. Why it matters: Advertisers running prediction-market-adjacent campaigns in Michigan or New York need to pull or geo-fence those campaigns now, since Google will reject them regardless of federal registration status.Wed, 15 Jul 2026 00:00:00 GMThttps://brandonlazovic.dev/pulse/2026-07-15/#chatgpt-ads-location-audience-exclusionhttps://brandonlazovic.dev/pulse/2026-07-15/#chatgpt-ads-location-audience-exclusionSearch Engine Roundtable spotted two new controls in ChatGPT Ads Manager: a location-exclusion setting that stops a campaign from serving to users in specified places, and an audience-exclusion setting that filters out selected custom audiences. OpenAI's own Help Center separately documents the audience-exclusion mechanism, requiring each excluded list to include at least 25,000 matched users. OpenAI has not yet published help documentation specifically describing the location-exclusion control. Why it matters: Brands running ChatGPT Ads campaigns can now suppress delivery to existing customers or ineligible regions, a targeting precision paid search teams have had on Google and Meta for years.Wed, 15 Jul 2026 00:00:00 GMThttps://brandonlazovic.dev/pulse/2026-07-15/#google-shopping-hide-sponsored-testhttps://brandonlazovic.dev/pulse/2026-07-15/#google-shopping-hide-sponsored-testGoogle's Ads Liaison Ginny Marvin confirmed the company is testing a hide and show sponsored results toggle inside Google Shopping, after a screenshot from Sachin Patel surfaced the option on X. Marvin said the test is about making ad labeling more consistent across search and shopping surfaces. Google shipped the equivalent hide and show toggle for standard search ads last October. This test extends the same interaction to Shopping listings. Why it matters: If this ships, sponsored Shopping listings could become easier for shoppers to collapse or hide, changing how much visibility paid placements get against organic results.Wed, 15 Jul 2026 00:00:00 GMThttps://brandonlazovic.dev/pulse/2026-07-14/#google-images-ai-overviews-image-generationhttps://brandonlazovic.dev/pulse/2026-07-14/#google-images-ai-overviews-image-generationMarking Google Images' 25th anniversary, Google is rolling out image generation inside AI Overviews: users can turn a text prompt into a custom image directly in Search results using the Nano Banana model. Google is also replacing the classic Google Images grid with a dynamic, personalized gallery with a Collections feature for saving images. Both features begin rolling out in the coming weeks, starting on U.S. desktop in English, and require a signed-in Google Account. Why it matters: Image generation inside AI Overviews adds another AI-native surface competing for the visual real estate that product and lifestyle photography used to own alone.Tue, 14 Jul 2026 00:00:00 GMThttps://brandonlazovic.dev/pulse/2026-07-14/#gemini-in-chrome-uk-expansionhttps://brandonlazovic.dev/pulse/2026-07-14/#gemini-in-chrome-uk-expansionGoogle is rolling out Gemini in Chrome to UK desktop users starting July 14, 2026, with an iOS expansion planned for next month. The assistant summarizes webpages, compares information across open tabs, and takes actions like drafting Gmail replies, checking Calendar, or pulling up Maps directions without leaving the page. It can also edit web images from a text prompt using Nano Banana 2, and Google says the models resist prompt injection and require confirmation before sensitive actions. Why it matters: An in-browser AI assistant that reads, summarizes, and acts across open tabs shifts more of the search-to-action journey away from the results page and into the browser itself.Tue, 14 Jul 2026 00:00:00 GMThttps://brandonlazovic.dev/pulse/2026-07-14/#gemini-spark-southeast-asia-languageshttps://brandonlazovic.dev/pulse/2026-07-14/#gemini-spark-southeast-asia-languagesGoogle is rolling out Gemini Spark, its AI content creation feature, in local Southeast Asian languages to Gemini Advanced (Ultra) subscribers the week of July 14, 2026, extending a tool that previously worked only in English. The rollout accompanies a regional usage report covering six countries, which found active users in Southeast Asia have more than doubled over the past year and that nearly 70 percent of prompts are now submitted in native languages rather than English. Why it matters: Native-language AI generation lowers the barrier to AI-first content creation in fast-growing Southeast Asian markets, a leading indicator for where AI-driven search behavior scales next.Tue, 14 Jul 2026 00:00:00 GMThttps://brandonlazovic.dev/pulse/2026-07-10/#shopify-hreflang-admin-togglehttps://brandonlazovic.dev/pulse/2026-07-10/#shopify-hreflang-admin-toggleShopify added a toggle under Online Store > Preferences > Social sharing and SEO that controls automatic hreflang generation. By default Shopify emits hreflang tags built from a store's Markets language and domain settings. Merchants who manage hreflang manually, through an app or edited theme code, can now disable the automatic tags so storefronts stop serving duplicate or conflicting hreflang signals. Why it matters: Duplicate hreflang from platform defaults fighting manual implementations is a recurring international SEO defect, and this toggle finally lets merchants pick one source of truth.Fri, 10 Jul 2026 00:00:00 GMThttps://brandonlazovic.dev/pulse/2026-07-10/#shopify-managed-markets-duties-inclusive-pricinghttps://brandonlazovic.dev/pulse/2026-07-10/#shopify-managed-markets-duties-inclusive-pricingShopify Managed Markets now calculates duties, import taxes, transaction fees, and currency conversion directly into the displayed product price for international buyers, instead of adding them as separate checkout line items. Merchants get consistent payouts through Shopify Payments, and Shopify absorbs any variance between the quoted duty amount and the actual customs charge. Merchants can preview localized price breakdowns using the platform's View as demo tool. Why it matters: Hidden fees at checkout are a well-documented cause of cross-border cart abandonment, so folding duties into the displayed price is a direct lever on international conversion for merchants selling across markets.Fri, 10 Jul 2026 00:00:00 GMThttps://brandonlazovic.dev/pulse/2026-07-10/#shopify-flow-copy-paste-workflow-stepshttps://brandonlazovic.dev/pulse/2026-07-10/#shopify-flow-copy-paste-workflow-stepsShopify Flow now lets merchants copy an action or condition step with Cmd/Ctrl+C and paste it with Cmd/Ctrl+V, within the same workflow or into a different one. Pasted steps retain their original configuration values and, for conditions, their logic, so they need only minor edits. The feature covers single steps only: triggers, bulk selections, and steps with custom configuration screens such as Send Marketing Email cannot be copied. Why it matters: Merchants who lean on Flow to automate recurring e-commerce operations, like feed syncs or redirect maintenance, can now duplicate proven step logic across workflows instead of rebuilding it by hand each time.Fri, 10 Jul 2026 00:00:00 GMThttps://brandonlazovic.dev/pulse/2026-07-09/#gpt-5-6-launchhttps://brandonlazovic.dev/pulse/2026-07-09/#gpt-5-6-launchOpenAI launched GPT-5.6 for general availability on July 9, 2026, following a limited preview, as a three-model family: flagship Sol, mid-tier Terra, and cost-efficient Luna. On Agents' Last Exam, a 55-field long-running-workflow benchmark, Sol scored 53.6, ahead of Claude Fable 5 by 13.1 points, and OpenAI says Terra and Luna beat Fable 5 at roughly one-sixteenth the cost. A new "ultra" setting coordinates multiple agents in parallel for the hardest tasks. Why it matters: This model's cost and reasoning benchmarks are now the baseline against which every "powered by GPT-5.6" product claim you see this week should be measured.Thu, 09 Jul 2026 00:00:00 GMThttps://brandonlazovic.dev/pulse/2026-07-09/#chatgpt-work-agenthttps://brandonlazovic.dev/pulse/2026-07-09/#chatgpt-work-agentOpenAI introduced ChatGPT Work on July 9, 2026, an agent built on GPT-5.6 and its Codex technology that gathers information across a user's apps and workflows to produce finished slides, sheets, docs, and web apps. It breaks complex projects into steps and can stay on a task for hours, including via Scheduled Tasks that turn new Slack or Microsoft Teams messages into updated documents while the user is away. OpenAI says more than 5 million people already use Codex weekly. Why it matters: An agent that turns raw inputs into finished deliverables unattended is a direct competitor to manual SEO and content-ops workflows, so teams should map which of their own recurring deliverables it could already produce.Thu, 09 Jul 2026 00:00:00 GMThttps://brandonlazovic.dev/pulse/2026-07-09/#alphaevolve-google-cloud-gahttps://brandonlazovic.dev/pulse/2026-07-09/#alphaevolve-google-cloud-gaGoogle made AlphaEvolve, its Gemini-powered code-optimization agent, generally available to all Google Cloud customers on the Gemini Enterprise Agent Platform on July 9, 2026. Users provide a baseline algorithm and a goal. AlphaEvolve then searches for improved, human-readable code instead of requiring a full rewrite. Since its December 2025 private preview, Google says BASF, JetBrains, and Kinaxis have used it to solve previously intractable business and research problems. Why it matters: For teams running large-scale ranking, feed, or crawl-optimization pipelines, a general-purpose code-optimization agent lowers the bar for squeezing performance out of existing algorithms without a full rewrite.Thu, 09 Jul 2026 00:00:00 GMThttps://brandonlazovic.dev/pulse/2026-07-09/#google-ads-ai-disclosure-panelhttps://brandonlazovic.dev/pulse/2026-07-09/#google-ads-ai-disclosure-panelGoogle launched a "How this ad was made" panel in My Ad Center, accessible globally across Search, YouTube, and Discover, that discloses when generative AI created or modified an ad, announced July 9, 2026. Ads made with Google's own AI tools are labeled automatically. Advertisers using outside AI tools can now manually disclose that use, and on-ad labels may appear depending on local rules. Google also points to its existing SynthID watermarking as a related safeguard. Why it matters: For advertisers and e-commerce marketers, AI-ad disclosure is now an account-level setting to configure, not just a content policy to read, so review My Ad Center before the labels start surfacing to shoppers.Thu, 09 Jul 2026 00:00:00 GMThttps://brandonlazovic.dev/pulse/2026-07-09/#meta-muse-spark-1-1-apihttps://brandonlazovic.dev/pulse/2026-07-09/#meta-muse-spark-1-1-apiMeta released Muse Spark 1.1 on July 9, 2026, a multimodal reasoning model upgrade with a public preview of Meta's Model API, the first time developers can access a Spark model programmatically. Meta says the model "excels at computer-use workflows" spanning multiple applications, choosing between writing automation scripts, direct interface clicks, or batched actions depending on the task. The company frames the release as a step toward agents that take action on a user's behalf. Why it matters: A third major lab now offers API-level computer-use agents alongside OpenAI and Google, so expect Meta-powered agentic traffic and automated interactions to start showing up in analytics alongside the others.Thu, 09 Jul 2026 00:00:00 GMThttps://brandonlazovic.dev/pulse/2026-07-09/#amazon-turnstile-token-capturehttps://brandonlazovic.dev/pulse/2026-07-09/#amazon-turnstile-token-captureAmazon Science published Turnstile on July 9, 2026, an open-source Rust proxy that sits between an agent harness and a model backend to record token IDs, log probabilities, and loss masks during agentic interactions, rather than relying on text transcripts. The tool targets "retokenization drift," where formatting changes make identical-looking transcript text map to different token IDs, which Amazon says degrades reinforcement-learning signals. It also merges multi-turn trajectories and supports mixture-of-experts routing and multimodal data. Why it matters: As more teams fine-tune or reinforcement-learn their own agents against real user interactions, capture-layer bugs like retokenization drift are the kind of silent data-quality issue that can quietly cap how well those agents learn.Thu, 09 Jul 2026 00:00:00 GMThttps://brandonlazovic.dev/pulse/2026-07-09/#mistral-studio-prompts-skills-recordhttps://brandonlazovic.dev/pulse/2026-07-09/#mistral-studio-prompts-skills-recordMistral introduced a system of record inside Mistral Studio on July 9, 2026, that versions, assigns ownership to, and makes traceable every prompt and skill an organization runs. Versions are immutable once deployed, changes carry audit trails and rollback, and Mistral says the feature connects prompts to real production behavior through observability and telemetry rather than acting as a static catalog. Non-engineering staff can now edit instructions directly while governance controls, including data residency, stay intact. Why it matters: Teams operationalizing LLM prompts for content, classification, or customer-facing workflows now have a concrete template for the change-control and audit trail that prompt sprawl otherwise makes impossible to reconstruct after the fact.Thu, 09 Jul 2026 00:00:00 GMThttps://brandonlazovic.dev/pulse/2026-07-08/#gpt-live-chatgpt-voicehttps://brandonlazovic.dev/pulse/2026-07-08/#gpt-live-chatgpt-voiceOpenAI launched GPT-Live on July 8, 2026, a full-duplex voice model that listens and speaks simultaneously instead of waiting for turns, and now powers ChatGPT Voice globally on iOS, Android, and the web. For complex questions, GPT-Live delegates to GPT-5.5 in the background while keeping the conversation flowing. GPT-Live-1 becomes the default for Go, Plus, and Pro users, GPT-Live-1 mini for Free users, with API access planned soon. Why it matters: As voice becomes a real interface for search and shopping help, ChatGPT's shift to full-duplex, delegation-based conversation raises the bar for how brands need to sound and respond when customers talk instead of type.Wed, 08 Jul 2026 00:00:00 GMThttps://brandonlazovic.dev/pulse/2026-07-08/#vllm-transformers-backend-speed-parityhttps://brandonlazovic.dev/pulse/2026-07-08/#vllm-transformers-backend-speed-parityHugging Face and vLLM announced on July 8, 2026, that the vLLM transformers modeling backend now matches or beats native vLLM throughput, closing a gap that previously forced teams to choose between an easy-to-modify model and a fast one. The team used static graph analysis and operation fusion to hit throughput parity across three Qwen3 model sizes, from a 4B dense model up to a 235B mixture-of-experts model on eight H100 GPUs. Teams activate it with a single flag. Why it matters: Cheaper, faster open-weight model serving lowers the cost of running the AI agents and retrieval systems that determine what content gets surfaced or cited, without locking teams into custom, hard-to-maintain inference code.Wed, 08 Jul 2026 00:00:00 GMThttps://brandonlazovic.dev/pulse/2026-07-07/#gemini-api-managed-agents-expansionhttps://brandonlazovic.dev/pulse/2026-07-07/#gemini-api-managed-agents-expansionGoogle announced four new capabilities for Managed Agents in the Gemini API on July 7, 2026. Background execution lets developers set background: true and poll a returned ID instead of holding an HTTP connection open for long tasks. Agents can now connect directly to remote MCP servers to reach private databases and internal APIs, and developers can mix in custom client-side functions alongside built-in sandbox tools. A credential-refresh option rotates access tokens without losing the sandbox's filesystem state. Why it matters: Async execution and direct MCP access make it more practical to wire Gemini agents into internal data systems that used to require custom polling infrastructure.Tue, 07 Jul 2026 00:00:00 GMThttps://brandonlazovic.dev/pulse/2026-07-07/#google-search-console-platform-propertieshttps://brandonlazovic.dev/pulse/2026-07-07/#google-search-console-platform-propertiesGoogle introduced platform properties in Search Console on July 7, 2026, a new property type that lets creators, including those without their own website, verify an Instagram, TikTok, X, or YouTube account and see how its content performs on Search and Discover. The property adds a Performance report for clicks and impressions by post and query, an Insights report for traffic trends and top posts, and Achievements for growth milestones. The rollout is gradual over the coming weeks. Why it matters: Off-site social and video content now gets the same query-level visibility data that on-site pages have long had in Search Console, extending measurable AI-visibility work beyond the owned website.Tue, 07 Jul 2026 00:00:00 GMThttps://brandonlazovic.dev/pulse/2026-07-07/#huggingface-foundry-managed-computehttps://brandonlazovic.dev/pulse/2026-07-07/#huggingface-foundry-managed-computeHugging Face and Microsoft announced Hugging Face models on Foundry at Microsoft Build 2026, a curated, weekly-refreshed catalog of open-weight models deployable in one click onto Foundry Managed Compute. Microsoft pre-stages model weights in Azure, builds and CVE-scans the runtimes (vLLM, SGLang, TensorRT-LLM, NIM, TEI, and llama.cpp), and applies the same enterprise security, governance, observability, and billing used across Foundry's other models. The catalog covers text, vision, audio, and multimodal models and is available now in preview. Why it matters: For teams running their own LLM pipelines, this removes the license-review, security-scanning, and GPU-sizing work that previously stood between a promising open-weight model and a production endpoint.Tue, 07 Jul 2026 00:00:00 GMThttps://brandonlazovic.dev/pulse/2026-07-07/#huggingface-skypilot-zero-egress-storagehttps://brandonlazovic.dev/pulse/2026-07-07/#huggingface-skypilot-zero-egress-storageHugging Face and SkyPilot jointly shipped zero-egress storage on July 7, 2026, adding Hugging Face Storage as a SkyPilot backend reachable through an hf:// URL. Because Hugging Face charges no egress or CDN fees, a SkyPilot job can mount a Hugging Face Bucket or Hub repo and read it from any of 20-plus supported clouds, Kubernetes, or on-prem clusters at no transfer cost. In Hugging Face's benchmark, a mounted model was training-ready in about 30 seconds on every cloud tested. Why it matters: Decoupling storage from GPU location removes the cross-cloud egress tax that has forced teams building large LLM pipelines to pin their runs to whichever provider holds a copy of the data.Tue, 07 Jul 2026 00:00:00 GMThttps://brandonlazovic.dev/pulse/2026-07-07/#shopify-payments-staff-permissionshttps://brandonlazovic.dev/pulse/2026-07-07/#shopify-payments-staff-permissionsShopify released four new staff permissions on July 7, 2026: manage payments settings, manage disputes, view payouts, and view tax documents. The permissions let store owners grant a team member only the specific financial access their role requires, instead of sharing full account capabilities. They are assigned per staff role from Settings, Users, Roles, Permissions, and Shopify says the rollout is already underway across all stores. Why it matters: Agencies and larger merchant teams can now limit who sees sensitive financial data without withholding the rest of the Shopify admin, closing a common over-permissioning gap.Tue, 07 Jul 2026 00:00:00 GMThttps://brandonlazovic.dev/pulse/2026-06-30/#nano-banana-2-lite-gemini-omni-flashhttps://brandonlazovic.dev/pulse/2026-06-30/#nano-banana-2-lite-gemini-omni-flashGoogle released two new generative models on June 30: Nano Banana 2 Lite, an image model that produces text-to-image outputs in four seconds at $0.034 per 1,000 images, and Gemini Omni Flash, which generates up to ten-second videos from text, image, or video inputs at $0.10 per second. Both ship across Google AI Studio, the Gemini API, and the Gemini Enterprise Agent Platform, and can be chained so an image from Nano Banana 2 Lite feeds Omni Flash for animation. Why it matters: Cheap, fast multimodal generation lowers the cost of producing on-page and ad creative at scale, which raises the bar for what counts as differentiated content.Tue, 30 Jun 2026 00:00:00 GMThttps://brandonlazovic.dev/pulse/2026-06-30/#shopify-inventory-transfer-metafieldshttps://brandonlazovic.dev/pulse/2026-06-30/#shopify-inventory-transfer-metafieldsShopify's June 30 changelog update lets merchants define and manage custom metafields on inventory transfers from the admin's Metafields and metaobjects settings, then add or edit values directly on the Transfer Create and Transfer Details pages. The Admin GraphQL API exposes the same fields programmatically for automation. Metafields copy over when a transfer is duplicated and can be used as analytics dimensions, supporting lot or serial tracking, logistics detail, and ERP or WMS synchronization. Why it matters: Structured transfer-level data gives merchants cleaner inventory provenance for feed accuracy and fulfillment reporting.Tue, 30 Jun 2026 00:00:00 GMThttps://brandonlazovic.dev/pulse/2026-06-30/#shopify-app-pixel-activity-loghttps://brandonlazovic.dev/pulse/2026-06-30/#shopify-app-pixel-activity-logShopify shipped an activity log on June 29 that tracks when and how a third-party app pixel's customer-data access has changed, available on each pixel's detail page inside the Customer events section of the admin. The log records every modification going back to June 3, 2026, giving merchants a full historical record of what data an app can see. It is available to all merchants running app pixels, with no separate setup required. Why it matters: Auditable data-access history helps merchants verify privacy compliance across every third-party tracking pixel on the storefront.Tue, 30 Jun 2026 00:00:00 GMThttps://brandonlazovic.dev/pulse/2026-06-30/#claude-sonnet-5-snowflake-cortex-aihttps://brandonlazovic.dev/pulse/2026-06-30/#claude-sonnet-5-snowflake-cortex-aiSnowflake announced on June 30 that Claude Sonnet 5 is available in private preview inside Snowflake Cortex AI, with Anthropic as launch partner. The model runs within the Snowflake security perimeter and plugs into CoCo, Cortex Agents, Cortex AI Functions, Cortex Inference, and Snowflake CoWork. Snowflake describes Sonnet 5 as approaching Opus-level quality on coding, debugging, analysis, and agentic tasks while running faster and at lower cost. Why it matters: Enterprise teams already inside Snowflake's data perimeter get a frontier coding and agent model without moving data outside it.Tue, 30 Jun 2026 00:00:00 GMThttps://brandonlazovic.dev/pulse/2026-06-30/#snowflake-cortex-sense-previewhttps://brandonlazovic.dev/pulse/2026-06-30/#snowflake-cortex-sense-previewSnowflake announced Cortex Sense on June 30, entering private preview in mid-July 2026. The system automatically builds semantic understanding of enterprise data from existing signals such as past analyst queries, transformation-tool models, and BI metrics, without manual curation, and flags gaps or conflicts for a human to resolve instead of guessing. In Snowflake's own testing, Cortex Sense raised agent accuracy on benchmark questions from 24.1% to 86.3% while cutting per-query cost from $1.76 to $0.59. Why it matters: Grounded semantic context for enterprise data matters because missing context, not weak models, is the usual reason warehouse-native agents give wrong answers.Tue, 30 Jun 2026 00:00:00 GMThttps://brandonlazovic.dev/pulse/2026-06-30/#bigquery-conversational-analytics-gahttps://brandonlazovic.dev/pulse/2026-06-30/#bigquery-conversational-analytics-gaGoogle announced general availability of Conversational Analytics in BigQuery on June 30, letting users query native BigQuery tables, Iceberg lakehouse tables, and cross-cloud sources like Databricks, AWS Glue, SAP, and Salesforce using natural language instead of SQL. The tool runs multi-step analyses with BigQuery AI functions for root-cause analysis, forecasting, and anomaly detection, shows its reasoning and generated SQL, and supports scheduled autonomous agents for proactive monitoring, all under BigQuery's existing governance and security controls. Why it matters: Natural-language, agent-run analysis at GA lowers the skill floor for ad hoc SEO and revenue analysis directly against warehouse data.Tue, 30 Jun 2026 00:00:00 GMT