đ” Notion's new Workers can build Stripe Dashboards and is a Claude Backlash brewing?
Plus: Google's vibe coded widgets, the companies who are ditching AI features, a genuinely useful bookmark tool you can use.
Hi product people đ,
This week, Notion unveiled a series of major new announcements as part of its Developer Platform that could have a significant impact on the product development process. Weâll take a look at how its new Workers and âFeature Creaturesâ can build internal tools and connect with third parties like Stripe.
Plus, Google announces the launch of agentic workflows and vibe coded widgets - could Apple follow suit? And Anthropicâs head of product reveals her vision for the future of product user experiences, but is a Claude backlash brewing?
Enjoy the rest of your weekend!
Rich
Watch on YouTube | Follow on Substack Notes
Key reads and resources for product teams
New from the Department of Product Substack this week:
Knowledge Series - What is the Command Line Interface and why do they matter to product teams?
OpenAIâs co-founder called them âsuper exciting,â and Spotifyâs CPO uses one every morning - but command line interfaces arenât what youâd expect. Once the exclusive domain of engineers, CLIs are experiencing a renaissance thanks to AI agents, which now prefer text-based interfaces to clicking buttons. Find out why product teams should care about this old technology in 2026, with practical examples from Google, Stripe, and Spotify.
New in the AI Library - Build an early warning system that flags emotional churn signals
Build a churn prediction dashboard that scores customer accounts by engagement health, using behavioural signals like feature usage, survey responses, and beta participation. Outputs a weekly ranked report of at-risk accounts with warning signals and suggested actions, filterable by segment, tier, and geography. (Department of Product)
Org design - How to run an AI-native engineering org
Fiona Fung, Director of Engineering for Claude Code, walks through what broke at Anthropic (review, ownership, hiring) and the norms we had to rewrite to keep shipping. (YouTube)
Design systems - Microsoftâs new âAI-forwardâ design system explored
Microsoft is redesigning how AI thinks alongside you. Jon Friedman, writing for Microsoft Design, explains the Copilot Design System - a new framework that treats AI as a true thought partner rather than a fragmented tool. (Microsoft Design)
Case study - How Slack rebuilt its notifications system
Slackâs notification system was broken which meant users couldnât understand their settings, mobile and desktop behaved differently, and confusion drove it into the top three customer complaints. Senior engineers Frances Coronel and Shilpa Kannan reveal how they rebuilt the entire system from scratch, unifying four conflicting models into one, and cut support tickets while increasing user control by 5x. (Slack Engineering)
Interview - Anthropicâs Head of Product on how future products will anticipate your needs
Anthropicâs product head Cat Wu predicts AI will soon anticipate your needs before you even realize them - a shift from todayâs chatbots to proactive agents that handle work automatically. Wu discusses how Anthropic plans to stay ahead of competitors and what managing âfleets of AI agentsâ really means for the future of work. (Anthropic)
Strategy - Is Software losing its head?
As software gets stripped down to its core, the rules for competitive advantage are being rewritten. Some will stay durable. Others will vanish. Read Andreessen Horowitzâs Seema Ambleâs full analysis on what changes when humans leave the building. (Andreessen Horowitz)
UX - The death of the empty state
Adi Leviim, a founder and engineer, argues AI products abandoned 40 years of HCI principles like signifiers and affordances for what amounts to no design at all. Read how Leviim envisions better first-run experiences using worked examples, starting verbs, and exposed model limits - design patterns that exist today but remain unused across the industry/ (UX Collective)
Analysis - Claude just overtook OpenAI - but is a backlash brewing?
Claude just overtook OpenAI as the most-used AI model in business - but the data comes with some serious caveats as the company risks a major backlash from the developer community.
New product features and innovation this week
Google held its Android event this week, where it released multi-step task automation capabilities for Gemini AI on Android devices. The feature lets users describe complex tasks through natural language, and Gemini executes them across multiple apps with user confirmation.
When you press your phoneâs power button and describe a task, Gemini reads your screenâs current content as context. It then performs sequential actions - like copying a grocery list from Notes, opening your shopping app, adding items to cart, and pausing before checkout. The AI also auto-browses the web to complete tasks like booking appointments, fills out forms using your saved Personal Intelligence data.
Another feature worth knowing about is is âvibe-codingâ for widgets. Users simply describe what they want in natural language (âSuggest three high-protein meal prep recipes weeklyâ) and Gemini automatically generates a custom widget. This could be the first step towards fully vibe coded apps that work natively in Android and the agent automations could change the way users interact with apps, but is iOS also set to adopt similar agentic features?
Apple has historically been openly hostile towards the idea of agents but it is reported to be considering rolling out similar features in upcoming versions of iOS. WWDC will reveal more about what Apple thinks - with potentially significant implications for product teams.
Notion builds a new Developer Platform and CLIs; Stripe live agentic dashboard impresses
Notion held its annual developer conference - with some powerful new features for product teams.
Theyâre branding this set of announcements the âNotion Developer Platformâ and the core concept at the heart of this is something they call Workers. These are small programs that Notion hosts and runs in its own environment with the pitch that developers and coding agents can write the logic and Notion takes care of everything else.
Workers can connect to any external system with an API - third party services like Stripe, Salesforce, and Snowflake, or internal databases and tools - and pull live data into Notion as a shared canvas.
In this example, a âStripe Syncâ worker was created in Notion to sync data from Stripe to Notion. The result was a live Notion database with Stripe customers, subscriptions, and invoices syncing in real time, which a personal agent then turned into a team dashboard:
They also showed something called âRenewal Ronnieâ, an agent built on top of that synced data that runs nightly, checks every customer renewal, and prompts the account team to reach out proactively.
Another demo showed a support team use case where a "Feature Creature" agent could check whether a feature flag was enabled for a customer, then attempt to toggle it - but only after validating the customer was on the right plan. When Gummy Bear Aeronautics turned out to be on Pro rather than Enterprise, the Worker blocked the toggle and logged the entire interaction to a shared customer database in Notion. Once built, the support team could run it themselves without involving engineering.

For product teams, this could be super-helpful for scenarios where you want to build your own internal tools or dashboards that proactively notify you of specific, relevant changes.
Microsoft, TikTok and Amazonâs AI feature cleanup
Feature bloat is a real risk in 2026, as AI makes it easier than ever for product teams to ship new features just because they can. This week, some companies started to slow down and assess and remove some AI features following user feedback.
TikTok is scaling back its AI summaries feature after complaints from users that it created bizarre and inaccurate captions and Microsoft is reported to be cutting back on the rollout of Copilot after Customers described sine Copilots as ranging from unnecessary to âfunctionally useless.â The Windows team has already removed Copilot entry points from Photos, Widgets, and Notepad. Gaming Copilot on Xbox is being wound down entirely.
Amazon is ditching its Rufus brand in favour of Alexa. Amazon says Alexa for Shopping brings together the Alexa+ and Rufus capabilities, signalling another step towards the simplification of pre-existing AI features.
Tools you can use
A1 MCP - Connect your AI coding tool to a curated library of real websites. Search by style, font, technology, and colour - without leaving your AI coding tool.
Dessn - a new design tool that runs directly in your codebase. Raised $6m this week.
Mark II - the bookmark reinvented. One of the few new âAI-poweredâ pieces of hardware that actually seems useful. Mark is a physical bookmark that tracks everything you read. The new version of Mark comes with a highlighter that lets you highlight passages and save them to apps like Apple Notes or Notion.
đ Product data and trends to stay informed
Microsoft has released its annual Work Trend Index report. The research analyzes how workers actually use AI agents and copilots in Microsoft 365, based on telemetry from over 100,000 conversations and a global survey of 20,000 AI users.
The analysis reveals that nearly half (49%) of Copilot conversations support cognitive work - analyzing information, solving problems, and thinking creatively. The remainder splits between working with people (19%), producing work (17%), and gathering information (15%). Within cognitive work, decision-making and problem-solving dominate at 28% of all usage.
âPodslopâ is the new term for AI generated posts. According to new analysis, 39% of new podcasts were likely AI generated.
LinkedIn is planning more layoffs along with operational changes. A new leaked memo from their CPO says âOur fastest moving teams are focused, have fewer layers, and leverage AI to move quickly⊠the changes weâre making across our Product orgs are based on creating more agile teams in this model.â
When tracking AIâs impact on jobs, donât just watch for mass layoffs. Pay attention to companies quietly cutting hiring plans. SaaS company Monday.com is an example - it went from planning 20% workforce growth to expecting headcount to stay âflatâ through 2026, reflecting the âproductivity gains AI is already deliveringâ across their organization.
This week, for the first time, Anthropic passed OpenAI in business adoption, according to the latest release of Ramp AI Index. Adoption of Anthropic rose 3.8% in April to 34.4% of businesses. OpenAI adoption fell 2.9% to 32.3%. Overall AI adoption rose 0.2 percentage points to 50.6%.
Website visits to Claude were up a massive 34% in April, while ChatGPT was down -3.84% and Grok down-14.4%.
Airbnbâs CEO Brian Chesky revealed that AI now writes 60% of Airbnbâs new code, with AI resolving 40% of customer support issues without human intervention. He also said managers who wonât get hands-on with tools like Claude Code âhave no placeâ at the company. Anthropicâs CEO says AI writes 90% of their code.
New analysis shows that over reliance on AI in software engineering is rotting engineersâ brains. And the former Adobe CPO argues that cognitive enhancement is better than cognitive offloading.
44% of DoP survey respondents said they prefer the idea of working with a team of AI agents over human colleagues. But for now at least, 56% say working with a team of AI Agents âsounds awfulâ:
Paid subscribers get the full DoP Substack including: The Knowledge Series for sharpening your tech / AI skills, the AI Prompt and Skills library and DoP Deep dive reports for in-depth analysis to learn lessons from the worldâs top tech companies.





The Notion Workers thing is interesting but the Stripe integration is where it gets real. No-code users building dashboards that pull live revenue data is a big shift. Question is whether the auth model holds up at scale.
The Stripe clone is the bait. The real wedge is all the internal software companies never build because itâs too small for engineering but too painful to keep manual. Thatâs where Notion Workers could matter: renewal checks, access requests, support exceptions, customer health workflows, sales handoffs.
The big question is whether Notion can make these workflows governed enough to trust. If no, every team just creates its own mini shadow-IT stack.