Product Intelligence
Decide what to build next, powered by Product Intelligence
Use what leading product teams launched, priced and built for themselves, and what happened next, to shape your own product. More than 20,000 judged signals become evidence for the decisions on your roadmap.
- 17 Mar 2026LaunchedV8 model to alpha testers, with better instruction-following
- 14 Apr 2026New versionV8.1, with V7-like aesthetics and better style references
- 11 Jun 2026Made defaultV8.1 becomes the default model for every generation
Learn from what world-leading product teams ship
A sample of recent signals · select one to open it
Filter by signal type, or by the team you want to learn from. A Perspective carries the author’s name and role, with product leaders prioritised to help shape your own product perspective.
product signals judged and analysed for product relevance
strategic decisions tracked, the best written up as case studies
product leaders and their perspectives tracked
CursorLaunchLaunched Origin, code hosting built into the editor with agentic workflowsLaunchA leading team has put something genuinely new into the world. See how they scoped it and shipped it while it is still fresh enough to learn from.
LovableFeatureAdded a /goal command for up-to-10-hour uninterrupted builds without user promptsFeatureThe everyday craft of shipping — how strong teams extend a product once it is live. The closest thing to watching someone else run your roadmap well.
DoorDashInternal toolBuilt Flux, a cloud platform automating 130,000 engineering tasks a monthInternal toolWhat a team built for itself rather than for customers. Nobody puts internal tooling on a roadmap, so this is the hardest thing to find out and the closest you get to how they really work.
StripeAI adoptionInternal coding agents now merge over 1,000 pull requests every weekAI adoptionA company changing how it works with AI, not what it sells. Borrow the techniques that are already working somewhere before you spend a quarter proving them yourself.
AppleResearchInternalized Visual Thinking reduces inference overhead in multimodal video reasoningResearchA technique, architecture or benchmark a company published. What was learned rather than what shipped — usually six months ahead of the product that will use it.
AirbnbOrgHiring a Senior Software Engineer (AI/ML) into Trust, based in BangaloreOrgHires, departures, reorganisations, a new team, an acquisition. Where a company puts its people is the earliest readable signal of where its strategy is going.
AnthropicEarningsQ2 revenue doubles to $11.6B, outpacing OpenAI’s $6.7B for the quarterEarningsWhat is actually working commercially, in their own numbers. The check on every other signal: plenty gets shipped, far less makes money.
LinearPerspectiveHead of Product questions the premise that AI agents lack adoptionPerspectiveA point of view, a lesson or a public stance from a named product leader. Borrow the judgement of people already doing your job well, with their role attached so you can weigh it.Eight signal types, labelled before they reach you. Filter to the kind of thing the decision in front of you actually needs — pricing moves when you are repricing, internal tooling when you are planning a build.
How it works
Judged through a product lens before it ever reaches you
Most feeds hand you everything and leave the reading to you. This one asks a single question of every item first — do product teams need to know about this? — and only what survives gets written up.
The filter is the product. Most of what is collected is thrown away, and what survives is rewritten so you can tell in one line whether it matters to you.
The people
See how top product leaders made the call you’re facing
Before you decide, find out what leaders who faced the same question chose, and why. The platform follows 500+ product leaders across posts, talks, podcasts and earnings calls, so when you’re weighing usage-based pricing, an AI agent or a flatter org, you see who has done it, what they said they learned and where their views differ, with every view linked to the original.
Nan YuHead of ProductLinear
Yuhki YamashitaChief Product OfficerFigma
Geoff CharlesChief Product OfficerRamp
Tom OcchinoChief Product OfficerVercel
Nick TurleyHead of ChatGPTOpenAI
Mike KriegerCo-lead, Anthropic LabsAnthropic
Boris ChernyClaude CodeAnthropic
Will GaybrickPresident, ProductStripe
Thariq ShihiparMember of Technical StaffAnthropic
Dharmesh ShahCo-founder & CTOHubSpot
Severin HackerCo-founder & CTODuolingo
Alexandr WangChief AI OfficerMeta
Demis HassabisCEO, Google DeepMindGoogle
Andy JassyPresident & CEOAmazon
Amjad MasadCo-founder & CEOReplit
Aravind SrinivasCo-founder & CEOPerplexity
Brian CheskyCo-founder & CEOAirbnb
Marc BenioffChair & CEOSalesforce
Alex BouazizCEO & Co-founderDeel
Brian ArmstrongCo-founder & CEOCoinbase
Clement DelangueCo-founder & CEOHugging Face
Dario AmodeiCo-founder & CEOAnthropic
David BaszuckiFounder & CEORoblox
Dara KhosrowshahiCEOUber
Drew HoustonFounder & Co-CEODropbox
Pavel DurovCo-founder & CEOTelegram
Matthew PrinceCo-founder & CEOCloudflare
Eric GlymanCo-founder & Co-CEORamp
Daniel EkExec Chair & Co-founderSpotify
Evan SpiegelCo-founder & CEOSnapchat
Greg BrockmanPresident & Co-founderOpenAI
Howie LiuCo-founder & CEOAirtable
Ivan ZhaoCo-founder & CEONotion
Jack DorseyBlock Head & CEOBlock
Josh ReevesCo-founder & CEOGusto
Luis von AhnCo-founder & CEODuolingo
Neal MohanCEOYouTube
Parker ConradCo-founder & CEORippling
Patrick CollisonCo-founder & CEOStripe
Guillermo RauchCEOVercel
Sam AltmanCEOOpenAI
Satya NadellaChairman & CEOMicrosoft
Sundar PichaiCEO, Google & AlphabetGoogle
Tony XuCo-founder & CEODoorDash
Tim CookCEOApple
Tobi LütkeFounder & CEOShopify
Vlad TenevChairman & CEORobinhood
Wade FosterCo-founder & CEOZapier
Zach PerretCo-founder & CEOPlaid
Dylan FieldCo-founder & CEOFigma513tracked in totalTheir judgement, applied to your decision. Ask the Thought Partner “how are product leaders handling this?” and the answer names who said what, with their role and the source, so you can weigh it against your own context.
Internal tools & AI adoption
Internal tools and AI adoption processes you can draw inspiration from
See what the best product teams have built for themselves — an agent that writes a third of the code, a sandbox where agents can run safely, a pipeline that made inference affordable. Each one is written up with what was built, the numbers behind it, and what a product team should take from it, so you can shape your own internal tooling and ways of working on patterns that are already running in production somewhere rather than designing them from scratch.
Enough detail to scope your own version. Open any of them for what was built, the numbers behind it, and what a product team should take from it — enough to brief your own team, size the equivalent for your stack, or decide it is not worth doing.
Deep Research
Get a week of product research done in minutes
Ask a broad question and Deep Research works through everything tracked to answer it — running its own searches, following what it finds through as many rounds as the question needs, then writing one report with every source attached. Start from the library of 51 prompts or write your own.
General-purpose deep research
Perplexity
ChatGPT
Gemini
ClaudeSearches the open web
Sources are whatever ranked this morning. Nothing has been checked for whether a product team needed to know about it, so the filtering is left to you — after the answer, not before it.
Department of Product
Deep ResearchSearches signals already judged for product relevance
Every item was read and judged against one question before it was ever stored — do product teams need to know about this? The filtering happened first, so the report is built from evidence that already earned its place.
It can still reach the open web — capped, and only when a question needs something the tracked set does not hold. When it does, the report names it as a web source in the sentence and says it has not been through the platform’s classification, rather than quietly mixing it in with the rest.
Bundling AI into an existing tier is the most common approach across the digitised pricing pages, with usage metering close behind. Almost nobody charges for AI as a standalone product.
The companies that moved to usage almost all kept a seat floor underneath it. Pure usage makes revenue unforecastable for the buyer as well as for you, and procurement punishes that harder than a high price.
If your cost per active user is unbounded, a flat tier prices for the median and loses money on exactly the users you most want to keep.
A week of desk research, done while you make coffee. It keeps searching until it has enough rather than answering from the first page of results, so what comes back is a structured report you could take into a pricing meeting — with the evidence behind every line.
Sample reports
Briefings
Your own research briefings, in your team’s inbox
Turn any Deep Research question into a briefing. Choose how often it runs, weekly, fortnightly or monthly, and who in your workspace gets it, and every send is a fresh report on what has changed, with every source linked.
Write the question once. It runs as written on every send and arrives at midday, UK time, so the research you keep meaning to do turns up whether anyone remembers or not. Start from your own question, a prompt from the library or one of the examples below.
Example briefings
See more of the platform