The AI Briefing

This week Apple opened Siri AI to every app developer, so any app can now hand its actions to Siri and let people use it without ever opening it. I walk through how that works with a food delivery example, why Uber’s agents arranged just four rides last quarter, and whether product teams will be happy for their apps to become invisible.In this video I also look at OpenAI’s GPT-Live voice agents making phone calls, how OpenAI used GPT-6 Astra to design a physical device, Perplexity running Meta ads for a lobster seller, and how Klaviyo, Linear, Docker, Deel and Sentry are handling agents at work. Plus the latest data on Anthropic’s amended IPO filing, data centre votes in the US elections, Sierra’s growth, and Reddit and AI search, plus how Google is fighting AI phone scams.➡️ Subscribe to Department of Product on Substack: https://departmentofproduct.substack.com/LinksSiri AI is now open to all developers (Apple Developer)Amazon Shop Direct Merchant Workspace (Amazon)Tomo pays with Shop Pay (Tobi Lütke)Uber only fulfilled 4 rides from AI agents last quarter (The Verge)Rafael K. Streit’s Siri AI demo with TripsyGPT-Live-1 is now in the API (OpenAI)OpenAI is merging ChatGPT’s Chat and Work tabs (TechCrunch)Accelerating engineering for the physical world (OpenAI)How Get Maine Lobster 3X’d Meta ad revenue with Perplexity (Perplexity)How Klaviyo scaled analytics by getting out of the way (Klaviyo Engineering)Our organizations are being DDoSed from the inside (Karri Saarinen)The CVE that did not exist (Docker)How Deel runs its agent platform (Deel)David Cramer on AI in interviewsAnthropic’s amended IPO filing (Business Insider)Voters deliver an “AI earthquake” (Axios)Sierra: our customers’ ambition is accelerating our growthReddit plunges after a weak outlook (CNBC)Reddit’s CEO warns Google’s “AI Mode does not drive traffic” (Department of Product, The Product Briefing)Which types of mentions correlate with AI visibility? (Ahrefs)Google’s fight against APAC scams (Google)

This week Ramp’s co-CEO Eric Glyman said non-engineers wrote 7,000 of Ramp’s pull requests in a single quarter, most of them from customer experience. Replit, Uber and Meta are seeing the same shift, and Ramp and PostHog both say design is struggling to keep up. Google’s Gemini 4 Carbon went straight to the top of the Intelligence Index, and Perplexity added a voice mode that can run its web app for you.In this video I look at what happens when anyone can ship a fix, the week’s launches from Canva, Hex, Perplexity and Google DeepMind, why more products now pick a model for each task, and what Stripe’s Link balance means for SaaS pricing when agents start doing the buying. Plus the latest data on Anthropic’s IPO filing, Azure, AWS capacity and AI’s return on investment, and how AI is bringing ultrasound scans to mothers who’ve never had one.➡️ Subscribe to Department of Product on Substack: https://departmentofproduct.substack.com/🔗 LINKSLive Voice Blueprints: 5 Features You Can Build with GPT-Live, Gemini Live and more (Department of Product Knowledge Series): https://departmentofproduct.substack.com/p/live-voice-blueprints-5-featuresEric Glyman on Ramp’s 7,000 non-engineer pull requests: https://x.com/eglyman/status/2115478868060504404Why Ramp is worth $60bn, with Eric Glyman (Brainfood): https://www.youtube.com/watch?v=BqMcJyCVmNoHamid Dadkhah on AI’s boost to engineering, product and design: https://x.com/diegozaks/status/2113331739263115743Tim Glaser on PostHog’s exit feedback: https://x.com/timgl/status/2113324756992999637Meta Q3 2026 follow-up call transcript: https://s21.q4cdn.com/399680738/files/content_files/META-Q3-2026-Follow-Up-Call-Transcript.pdfIntroducing Canva Craft (Canva): https://www.canva.com/newsroom/news/canva-craft/A new era for Hex, and data (Hex): https://hex.tech/blog/hex-new-era/Perplexity voice mode on the web (Aravind Srinivas): https://x.com/AravSrinivas/status/2115555247217320196Robbie Petersen on dopamine and cortisol interfaces: https://x.com/robbiepetersen_/status/2114083326604742913Gemini 4 Carbon: our next frontier model (Google DeepMind): https://deepmind.google/blog/gemini-4-carbon-our-next-frontier-model/GitHub Copilot Fusion (Kyle Daigle): https://x.com/kdaigle/status/2115483846838034523OpenAI’s Decisions API goes wide (TechCrunch): https://techcrunch.com/2026/10/22/openai-makes-its-jev-competitor-widely-available-as-decision-model-competition-heats-up/Introducing Link balance (Link): https://x.com/link/status/2114799287645393098Thais Castello Branco on agent budgets and pricing: https://x.com/thaiscbranco_/status/2114813240182251791Rami Banna on Stripe Projects (WorkOS init 2026): https://www.youtube.com/watch?v=YP6misI1xXwJeff Weinstein on pay-per-use for agents: https://x.com/jeff_weinstein/status/2089755499806031999Anthropic’s S-1 filing (SEC): https://www.sec.gov/Archives/edgar/data/2130357/000162828026067168/anthropicpbc-sx1.htmMicrosoft shares Azure’s revenue in dollars (TechCrunch): https://techcrunch.com/2026/10/28/microsoft-just-shared-azures-actual-revenue-in-dollars-33-2-billion/AWS largely sold out of capacity into 2028, Jassy says (The Information): https://www.theinformation.com/briefings/aws-sold-capacity-2028-jassy-says1Password surpasses $500 million in revenue as AI agents drive growth (The Wall Street Journal): https://www.wsj.com/tech/ai/1password-surpasses-500-million-in-revenue-as-ai-agents-drive-growth-903ccd4bVercel’s survey of SaaS authentication: https://x.com/vercel/status/2113397272943976560The Work AI Index 2026 (Glean): https://www.glean.com/work-ai-institute/work-ai-indexEmpowering frontline maternal care with AI-assisted ultrasound (Google Research): https://research.google/blog/empowering-frontline-maternal-care-with-ai-assisted-ultrasound/

This week Anthropic opened a research preview of Claude Duet, a proactive agent that sends you a morning brief, checks your apps every few hours and only messages you when something needs you. Google launched Universal Cart, Stripe says agent purchases through Link rose 38 times in a month, and Replit made OpenAI its default model over Claude.In this video I look at why ambient, proactive agents might be the next big shift in UX, how agents are starting to spend real money, the debate over whether AI features should be cheap, and the latest data and trends, from AI agents doing real science to AI finding security holes faster than people can patch them. Plus the developer who used Claude Code to strip the ads off his parents’ TV.➡️ Subscribe to Department of Product on Substack: https://departmentofproduct.substack.com/🔗 LINKSWebMCP in Practice (Department of Product Knowledge Series): https://departmentofproduct.substack.com/p/webmcp-in-practiceIntroducing Claude Duet (Anthropic): https://claude.com/blog/claude-duetMike Krieger’s Claude Duet demo: https://x.com/mikeyk/status/2112986418503110772Claude Duet for Shopify (Harley Finkelstein): https://x.com/harleyf/status/2112991076520993155Stripe Link in Claude Duet (Jeff Weinstein): https://x.com/jeff_weinstein/status/2113001845803364629Arnab Bose, Asana CPO, on proactivity (MTS): https://x.com/MTSlive/status/2110510584563454385Anthropic launches Claude Duet (Business Insider): https://www.businessinsider.com/anthropic-answers-meta-openai-new-personal-agent-claude-duet-2026-10Using agentic pipelines to turn test coverage into a daily habit (Atlassian): https://www.atlassian.com/blog/bitbucket/using-agentic-pipelines-to-turn-test-coverage-into-daily-habitAn intelligent new shopping cart (Google): https://blog.google/products-and-platforms/products/shopping/universal-cart-launch/How AI assistants are spending real money (Upstarts Media): https://www.upstartsmedia.com/p/data-stripe-link-how-ai-assistants-spendMuse hits over 10M users (Business Insider): https://www.businessinsider.com/meta-q3-2026-earnings-mark-zuckerberg-muse-ai-app-users-2026-10Replit makes OpenAI its default over Anthropic (TechCrunch): https://techcrunch.com/2026/10/17/replit-is-no-longer-working-with-anthropic-ceo-says/Agents, rebuilt to do more for less (Notion): https://www.notion.so/releases/2026-10-19Spenser Skates on Ultramarine 1 Flash: https://x.com/spenserskates/status/2111302477694005593Parker Conrad on token costs: https://x.com/parkerconrad/status/2112694469137183005Sarah Sachs on token costs: https://x.com/sarahmsachs/status/2113063933296918620AI Overviews have nearly doubled across commercial searches (Semrush): https://www.semrush.com/blog/ai-overviews-commercial-modifiers-study/Using thousands of AI agents to uncover new leads in biological data (OpenAI): https://openai.com/index/scaling-ai-research-for-complex-biological-dataHow Claude designed protein binders against scorpion neurotoxins (Anthropic): https://www.anthropic.com/research/claude-scorpion-toxin-binder-designMass General Brigham’s FaceAge tool (Boston Globe): https://www.bostonglobe.com/2026/10/20/metro/ai-biological-age-face-selfie-free-tool/Patch the Planet (OpenAI): https://openai.com/index/coordinated-disclosure-of-critical-vulnerabilitiesMatthew Prince on php2shell: https://x.com/eastdakota/status/2113079352682570051Using Claude to reduce the attack surface of code (Anthropic): https://www.anthropic.com/research/attack-surface-reductionCleanYourTV (Gathin T.): https://github.com/Gathin23/cleanyourtvCHAPTERS00:00 Intro00:28 Claude Duet and the rise of ambient agents04:44 Agents are spending real money06:02 Loyalty to any one model is fading08:28 Data and trends this week11:50 And finally

Anthropic has published a report on how AI has changed work inside Anthropic itself, using its own hiring records, employee surveys, calendars, Slack and agent data, plus 17 interviews with staff. By August, the agents its employees ran were working 6 times as many hours as the employees themselves. I’ve read the whole thing and lined it up against what other companies have been saying and shipping, to see which parts are unique to Anthropic and which are already showing up everywhere else.In this video I look at why Anthropic is hiring far fewer juniors and how companies like Amplitude and Coinbase are getting flatter, how output is rising in sales, legal and finance as well as engineering, why human attention has become the bottleneck when agents do most of the work, and what Anthropic now looks for when it hires. Plus the three things it thinks employers should do about it.➡️ Subscribe to Department of Product on Substack: https://departmentofproduct.substack.com/🔗 LINKSProduct Briefing: Shopify’s CEO says “AI slop grenades” are ruining productivity: https://departmentofproduct.substack.com/p/shopifys-ceo-says-ai-slop-grenadesHow Employment is Changing at Anthropic (Anthropic): https://www.anthropic.com/research/employment-at-anthropic2026 AI Adoption Report (Wharton Human-AI Research and GBK Collective): https://knowledge.wharton.upenn.edu/special-report/2026-ai-adoption-report/Amplitude’s CEO on having fewer managers: https://x.com/spenserskates/status/2107992072322216156Coinbase’s President on flatter teams: https://x.com/emiliemc/status/2099582174693625926Gamma’s CEO on AI and junior talent: https://x.com/thisisgrantlee/status/2092971330727784844Codex adoption outside engineering (Greg Brockman, OpenAI): https://x.com/gdb/status/2091861755253371243Linear’s COO on non-engineering teams working like engineers: https://x.com/cjc/status/2098093185973317744Stripe’s Will Gaybrick on building more with agents (a16z): https://x.com/jeff_weinstein/status/2089523705986900278Thariq on slowing down with agents: https://x.com/trq212/status/2112219173288120531GitLab’s CEO on trusting code over producing it: https://x.com/bstaples/status/2092761154468851812Shane Parrish on “slop grenades”: https://x.com/c_valenzuelab/status/2100575631922614339The Thin Orchestrator (Meta Engineering): https://engineering.fb.com/2026/10/19/ml-applications/thin-orchestrator-ai-agent-architecture/Introducing Add to Slack (Slack): https://slack.com/blog/news/add-to-slackHeyGen’s AI interview round: https://x.com/wayne_liang_/status/2096762118075007114CHAPTERS00:00 Intro00:42 Fewer juniors, flatter teams03:32 Everyone starts building05:56 Human attention is the bottleneck08:45 Work in the open, hire for judgment11:05 What employers can doThumbnail: Dario Amodei, photo via People Matters. Chart from “How Employment is Changing at Anthropic” (Anthropic, October 2026), Figure 2.

This week Figma put real, running code on the design canvas with code layers, Google shipped Gemini 4 Argon and launched its Gemini agent for work, and new data from Ramp shows which AI tools companies are actually paying for.In this video I look at why Figma wants your prototypes to run on your real code, and how that compares with Coinbase letting agents tell its team what to build next. Plus Pinterest working out your “journeys” with a small AI model, why more of your users are about to be agents, the latest data and trends, and Baroness Campbell giving a speech in the House of Lords in her own AI voice.➡️ Subscribe to Department of Product on Substack: https://departmentofproduct.substack.com/🔗 LINKSDecision models explained (Department of Product Knowledge Series): https://departmentofproduct.substack.com/p/decision-models-explained-the-newDesign together with code layers (Figma): https://www.figma.com/blog/design-together-with-code-layers/Figma’s Chrome extension captures your website code (Figma): https://x.com/figma/status/2110603044346052714Here’s everything from Config India 2026 (Figma): https://www.figma.com/release-notes/?title=config-india-2026Payload is now in Figma Make (Dylan Field): https://x.com/zoink/status/2110609962540126655Karri Saarinen on what building teaches you: https://x.com/karrisaarinen/status/2108259688001490958Have your agent send us feedback (Brian Armstrong, Coinbase): https://x.com/brian_armstrong/status/2110562521786507570Google Cloud introduces the Gemini agent (Google): https://blog.google/innovation-and-ai/infrastructure-and-cloud/google-cloud/gemini-at-work/Gemini agents in your SCIM feed (WorkOS): https://workos.com/blog/gemini-agents-scim-non-human-usersGemini 4 Argon (Google DeepMind): https://deepmind.google/blog/introducing-gemini-4-argon/From activity to intent: generating user journeys with LLMs (Pinterest Engineering): https://medium.com/pinterest-engineering/from-activity-to-intent-generating-user-journeys-with-llms-9652b9417626YouTube Music launches new AI labels (TechCrunch): https://techcrunch.com/2026/10/14/youtube-music-launches-new-labels-as-ai-generated-music-floods-streaming-platforms/Vercel traffic stats (Guillermo Rauch): https://x.com/rauchg/status/2108733051283050964Rebuilding the web for agents (Siva Gurumurthy, Harvey): https://x.com/sgurumur/status/2110217735791579540Agents can now buy domains with the Vercel CLI (Vercel): https://vercel.com/changelog/agents-can-now-buy-domains-with-the-vercel-cliSentry’s headless signup experiment (David Cramer): https://x.com/zeeg/status/2110082396397920679AI category winners, from Ramp spend data (Leading Indicators): https://x.com/leadinginds/status/2110080629773635595Square’s new kiosk (The Verge): https://www.theverge.com/tech/1009086/square-kioskBaroness Campbell speaks with her AI voice in the House of Lords (ElevenLabs): https://x.com/ElevenLabs/status/2110037384192774210CHAPTERS00:00 Intro00:21 Figma brings real code to the canvas03:33 Features and launches this week06:37 More of your users are agents08:14 Data and trends this week10:30 And finallyThumbnail photo credits:Dylan Field: TechCrunch, CC BY 2.0, via Wikimedia Commons (Dylan Field TechCrunch Disrupt 2022)Screenshot: Figma, code layers demo

This week Meta released Muse Wonder, a new model built for one job: running your personal agent. Over the weekend Elon Musk shared a video of someone using Grok Bot to shop on Amazon from the back of a Tesla robotaxi. The Agent Wars are officially here.In this video I look at how Meta went from being written off in AI to one of the fastest-growing apps in AI, why Elon Musk is happy for Grok Bot to run on Claude, and what happens when all of these agents turn up at businesses that never invited them, from Amazon blocking Muse to Shopify and Sierra building a proper front door for agents. Plus four things product teams should be thinking about right now.➡️ Subscribe to Department of Product on Substack: https://departmentofproduct.substack.com/🔗 LINKSIntroducing Personal Agent Protocol (Sierra): https://sierra.ai/blog/introducing-personal-agent-protocolIntroducing Muse Wonder (Meta): https://x.com/AIatMeta/status/2109691148943999167Grok Bot in a Tesla Cybercab (Sawyer Merritt): https://x.com/SawyerMerritt/status/2109490463635914867Muse partners with PayPal (Alexandr Wang): https://x.com/alexandr_wang/status/2102418455622508722Muse for Small Business (Meta): https://about.fb.com/news/2026/09/introducing-muse-small-business/Meta made a Tamagotchi-like wearable for its Muse AI agent (TechCrunch): https://techcrunch.com/2026/09/23/meta-made-a-tamagotchi-like-wearable-for-its-muse-ai-agent/Muse has a serious 0-day (Ars Technica): https://arstechnica.com/security/2026/09/muse-metas-extraordinarily-privileged-ai-assistant-has-a-serious-0-day/Meta’s Muse AI sent a YouTuber’s address to a stranger (The Verge): https://www.theverge.com/ai-artificial-intelligence/1001886/meta-muse-ai-facebook-marketplace-security-concernsMuse creates detailed profiles of all your friends and family (Wired): https://www.wired.com/story/muse-creates-detailed-profiles-of-all-your-friends-and-family/Introducing dots, always-on agents (OpenAI): https://www.youtube.com/watch?v=uXspbC2srEQGrok Bot tops 400,000 users after its first month (Bloomberg): https://www.bloomberg.com/news/articles/2026-09-22/spacexai-s-grok-bot-agent-tops-400-000-users-after-first-monthAn AI phone support line built in one day (Eric Bandholz): https://x.com/bandholz/status/2108634812437156129Grok Bot will use the best back end model for each task (Elon Musk): https://x.com/elonmusk/status/2107724314451878104Amazon blocks Meta’s Muse AI agent (The Verge): https://www.theverge.com/tech/998078/amazon-blocks-meta-muse-ai-agent-shoppingNearly every shopping site is blocking Instinct and Muse (Matthew Sigel): https://x.com/matthew_sigel/status/2109043587174338937“Agentic commerce is a mirage” (Eric Seufert): https://x.com/eric_seufert/status/2109336052565782608Caller ID in the age of agents (Sierra): https://sierra.ai/blog/caller-id-in-the-age-of-agentsDefining how AI agents interact with your business (ElevenLabs): https://elevenlabs.io/blog/synthetic-voice-detection-and-personal-agent-protocolShopify partners with Muse for Shop Pay checkout (Tobi Lütke): https://x.com/tobi/status/2102090718546198790Nubank agents inside Muse and Grok (David Vélez): https://x.com/sytaylor/status/2109579345316512240Patrick Collison on agents in the economy: https://x.com/patrickc/status/2108706907859116280Substack - The new UX of AI Agents: https://departmentofproduct.substack.com/p/the-new-ux-of-ai-agentsCHAPTERS00:00 Intro00:39 Meta’s comeback03:45 Grok Bot wants to be everywhere06:04 The businesses on the other end09:14 What this means for product teams

The State of AI Report 2026 from Nathan Benaich at Air Street Capital came out this week, all 244 pages of it. I’ve read the whole thing and lined its biggest themes up against what companies have actually been shipping, to see how they’re playing out in the real world.In this video I look at agents moving out of engineering and into legal, support and accounting, why the software around a model now matters more than the model itself, how the best teams are getting their AI to improve from real usage, and the agent safety incidents that have product teams building identity, approvals and limits into their products. Plus how the report scored its own predictions from last year, and the ones for next year that stood out to me.➡️ Subscribe to Department of Product on Substack: https://departmentofproduct.substack.com/🔗 LINKSState of AI Report 2026 (Air Street Capital): https://www.stateof.ai/Building Uber’s legal redlining agent (Uber): https://www.uber.com/us/en/blog/building-ubers-redlining-agent/Zendesk custom agents pass one million runs in early access: https://x.com/TomEggemeier/status/2108284761525342347Gusto launches an assistant for accountants: https://x.com/edawerd/status/2107982627017990207Meta and Microsoft wean staff off Anthropic’s Claude (The Information): https://www.theinformation.com/articles/meta-microsoft-work-wean-staff-anthropics-claudeLinear’s COO on app vs MCP usage: https://x.com/cjc/status/2107584490961080680Introducing Microsoft-Decision-1: https://x.com/satyanadella/status/2108627923888754862Snowflake Decision (Snowflake): https://www.snowflake.com/content/snowflake-site/global/en/blog/snowflake-decision-cortex-ai-functionsClef-omni and a cheaper Clef-flash (Cloudflare): https://blog.cloudflare.com/clef-faster-cheaper-multimodal/Liquid AI’s d1 on AI Gateway (Vercel): https://vercel.com/changelog/liquid-ai-d1-is-available-on-ai-gatewayAmplitude tests Gemini 3 Flash in place of Sonnet 4.6: https://x.com/spenserskates/status/2108300750053147061How Hex used evals to improve an AI feature: https://hex.tech/blog/we-used-evals-to-improve-ai-feature/Self-healing Next Gen Zaps (Zapier): https://x.com/wadefoster/status/2107454444631646588Investigating unintended model actions (Anthropic): https://www.anthropic.com/research/investigating-unintended-model-actionsIntroducing Agent SSO (Okta): https://x.com/okta/status/2107909146351669528Agents can now buy domains with the Vercel CLI: https://vercel.com/changelog/agents-can-now-buy-domains-with-the-vercel-cliIntroducing Strands Box (AWS): https://aws.amazon.com/blogs/opensource/introducing-strands-box-ai-agent-sandboxes-powered-by-dogwood/Shopify shares compare-at prices with agentic channels: https://changelog.shopify.com/posts/compare-at-prices-now-displayed-to-agentic-channelsStripe’s retail partners on the agent protocol: https://x.com/stripe/status/2107533255646089319TikTok rolls out an AI shopping assistant and one-click checkout (TechCrunch): https://techcrunch.com/2026/10/05/tiktok-rolls-out-an-ai-shopping-assistant-and-one-click-checkout/Substack - How to manage AI agents effectively: https://departmentofproduct.substack.com/p/how-to-manage-ai-agents-effectivelyCHAPTERS00:00 Intro00:35 Agents leave the developer’s desk03:21 The harness is the product05:46 Agents that learn on the job07:50 Agents go off script10:11 The report marks its own homeworkThumbnail: Dario Amodei, frame from “Code with Claude 2026: A conversation with Dario Amodei & Daniela Amodei” (Anthropic, YouTube), upscaled. Slide from the State of AI Report 2026, Air Street Capital (stateof.ai), p.107.

Anthropic's Claude Artifacts just got a major upgrade, Spotify launched a conversational assistant that goes beyond song requests, and Apple filed a lawsuit against OpenAI with some wild claims. In this video I break down what Artifacts can now do with MCP connectors, how Spotify's new feature fits the broader shift toward conversational discovery, and what's actually being alleged in the Apple v OpenAI case.➡️ Subscribe to Department of Product on Substack: https://departmentofproduct.substack.com/🔗 LINKSClaude Artifacts MCP connectors announcement: https://x.com/ClaudeDevs/status/2077489907350856038Claude Code engineer on building dashboards with Artifacts: https://x.com/trq212/status/2076790799011131735"The Unreasonable Effectiveness of HTML": https://claude.com/blog/using-claude-code-the-unreasonable-effectiveness-of-htmlSpotify's conversational feature announcement: https://newsroom.spotify.com/2026-07-14/talk-to-spotify-announcement-beta/Apple v OpenAI lawsuit, wildest claims (The Verge): https://www.theverge.com/tech/964843/apple-openai-lawsuit-wildest-claimsFull complaint document: https://www.documentcloud.org/documents/28453238-complaint-for-trade-secret-misappropriation-and-breach-of-contract/OpenAI's Codex Micro keypad: https://openai.com/supply/co-lab/work-louder/ChatGPT vs Gemini market share data (Similarweb): https://x.com/Similarweb/status/2077322771260957122Google Images redesign (TechCrunch): https://techcrunch.com/2026/07/14/google-images-gets-a-pinterest-like-redesign-focused-on-discovery/Google Images 25th anniversary announcement: https://blog.google/products-and-platforms/products/search/google-images-25th-anniversary/Google connected apps, Instacart integration: https://blog.google/products-and-platforms/products/search/connected-apps/Starbucks building in-house software to replace Microsoft/IBM tools (Forbes): https://www.forbes.com/sites/sandycarter/2026/07/12/starbucks-just-fired-a-warning-shot-at-microsoft-and-ibm-ai-apps/Substack Deep Dive - 30+ examples of AI Assistants in products: https://departmentofproduct.substack.com/p/the-new-powers-of-ai-assistants-exploredCHAPTERS00:22 Claude Artifacts Get MCP03:45 OpenAI Smart Speaker Rumors07:11 Google Images Discovery Redesign

OpenAI just released a wave of new products: ChatGPT Work, GPT-Live, and a new model family (GPT-5.6). In this video we break down what each one actually does, how the ChatGPT Work experience compares to Claude's desktop app, and whether any of this is enough to shift the balance in the enterprise AI race against Anthropic.➡️ Subscribe to Department of Product on Substack: https://departmentofproduct.substack.com/🔗 LINKSGPT-5.6 announcement: https://openai.com/index/gpt-5-6/OpenAI on coding evaluations: https://openai.com/index/separating-signal-from-noise-coding-evaluations/#misleading-promptArtificial Analysis Intelligence Index: https://artificialanalysis.ai/evaluations/artificial-analysis-intelligence-indexMatei Zaharia on benchmarking: https://x.com/matei_zaharia/status/2075002190843011199Databricks on coding agent benchmarks: https://www.databricks.com/blog/benchmarking-coding-agents-databricks-multi-million-line-codebaseGPT-Live announcement: https://openai.com/index/introducing-gpt-live/Ramp enterprise AI adoption data: https://x.com/arakharazian/status/2075290495924715996Fidji Simo steps down as CEO of Apps: https://www.theverge.com/ai-artificial-intelligence/963738/openai-fidji-simo-steps-down-ceo-advisorGemini consumer growth data: https://x.com/Similarweb/status/2054821737062154317Ramp AI Index, July 2026: https://ramp.com/data/ai-index-july-2026Building the ultimate product dashboard:https://departmentofproduct.substack.com/p/how-to-build-the-ultimate-productCHAPTERS00:00 OpenAI's Big Week00:20 ChatGPT Desktop App05:21 GPT Live Voice07:04 Enterprise Battle

This week: Anthropic's Claude Fable 5 arrives as the first public Mythos-based model, AI agents get the ability to spend your money, and Amplitude bets on a future where products improve themselves.We cover how Fable 5's classifier system routes risky queries to a safer fallback model, what people are already building with it (games, video edits, landing pages, API diagrams), and the key prompting principles from Anthropic's new guide - including why intent matters more than instructions, how to use parallel subagents, and why effort is now a dial.Then we get into agentic payments: Visa and OpenAI's new tokenized transaction partnership, Mastercard's Agent Pay for Machines infrastructure for machine-speed micropayments, and Coinbase's MCP server that lets agents rebalance your crypto portfolio and pay for research data automatically.We also look at Amplitude's Wave - an agent that runs the full product improvement loop from signal to ship to measure - and ask whether self-improving products are the next frontier, or just AI-optimized products without a soul.Finally, data from Lovable's Build Economy report (80% non-technical builders, 60% haven't made money yet), Google's 1.2 million vibe-coded apps per week, HBR's finding that therapy and companionship now account for 11% of all AI use cases, and Atlassian research showing that admitting you use AI at work makes colleagues rate you 10x lazier.➡️Get more analysis on Substack: https://departmentofproduct.substack.com/ 🔗 LINKSFable 5Anthropic Fable 5 release: https://www.anthropic.com/news/claude-fable-5-mythos-5Anthropic Fable 5 prompting guide: https://platform.claude.com/docs/en/build-with-claude/prompt-engineering/prompting-claude-fable-5Minecraft roller coaster (single shot): https://x.com/JinjingLiang/status/2064437608495919138Table tennis game: https://x.com/McGreenBeats/status/2064746187354149316Video editing deck: https://thariqs.github.io/cc-video-editing-deck/Video editing thread: https://x.com/trq212/status/2064826394589442448Single-shot API architecture diagram: https://x.com/backnotprop/status/2064461973442707942Liquid glass SaaS landing page: https://x.com/alexprokhorov/status/2064467938522857562Agentic PaymentsVisa x OpenAI announcement: https://www.businesswire.com/news/home/20260610422687/en/Visa-Partners-with-OpenAI-to-Power-the-Next-Generation-of-AI-CommerceMastercard Agent Pay for Machines: https://www.mastercard.com/us/en/news-and-trends/press/2026/june/mastercard-launches-agent-pay-for-machines.htmlCoinbase MCP for agent trading: https://techcrunch.com/2026/06/11/coinbase-debuts-mcp-for-agent-trading/A guide to new Agentic Payment Abilities and Features https://departmentofproduct.substack.com/p/new-agentic-payment-abilities-andSelf-Improving ProductsAmplitude Wave: https://amplitude.com/waveJack Dorsey on company intelligence: https://block.xyz/inside/from-hierarchy-to-intelligenceData & TrendsLovable Build Economy report: https://thebuildeconomy.lovable.app/Logan Kilpatrick on Google app generation: https://x.com/OfficialLoganK/status/2064423388928790943HBR AI use case study: https://hbr.org/2026/06/how-people-are-really-using-ai-in-2026Atlassian research on AI disclosure: https://www.atlassian.com/blog/ai-at-work/new-research-shows-honesty-about-ai-use-at-work-is-backfiring00:00 Welcome and rundown00:35 Fable 5 04:35 Agentic payments updates06:05 Self improving products07:25 Data and trends roundup