Your team has a Definition of Done for a product increment. It has none for the 20-plus AI-supported outputs that leave the team each week: status reports, stakeholder emails, release notes, and updates for the C-level. Each one carries your team’s name. “I know quality when I see it” is the standard most teams actually run by, and you cannot audit it, teach it to a new colleague, or defend it when a claim turns out to be wrong. The AI Definition of Done fixes that with one page per task class, agreed by the team, before the output ships.
TL; DR: AI in Product 2026 — Food for Agile Thought #549
Welcome to the 549th edition of the Food for Agile Thought newsletter, shared with 35,498 peers. This week, Product Circle and Product Institute share the AI in Product 2026 survey show AI coding tools spreading faster than stronger operating models, while Elena Verna sees cheaper software creation opening a Mom-and-Pop SaaS lane for domain experts. Petra Wille counters AI possibilities with accountable product principles, and Sam McVeety and Amir Hormati tackle agent-ready context. Also, Isabel Juniewicz and Ed Zitron question whether increasing hyperscaler spending and the economics of generative AI can sustain the rush, or bubble?
Next, Janna Bastow warns that Slack loses product feedback once channels move on, and Sarah Guo argues that AI shifts durable advantage toward private data, judgment, and trust. Aakash Gupta and Rohan Varma push the logic further, describing AI-native teams that build before they coordinate as the AI way, while Matthew Hodgson adds that enterprises need persistent funding and governance to make AI product operating models work. Then, Gregor Ojstersek shows that top engineering teams are already reshaping structures around AI.
Lastly, Mark Graban warns that tone policing in teams drives bad news underground, while Barry O’Reilly argues that AI raises the premium on visible, codified judgment that requires transparency, not enforced harmony. Johanna Rothman and Sonya Siderova shift the focus from faster tasks to slower systems, where wait times and flow debt shape delivery. Finally, Matteo Tittarelli extends that logic to GTM, where context, skills, orchestration, and integrations must compound across cycles.
Your team ships AI outputs that nobody fully trusts; you needed to be quick, and “dirty” tagged along. That ungoverned automation becomes AI debt the moment a stakeholder asks who owns it. The AI Delegation Lifecycle turns six agile skills you already practice into six explicit decisions that govern delegated AI work and produce audit-ready evidence without a separate report.
TL; DR: ROT (Return on Tokens) — Food for Agile Thought #548
Welcome to the 548th edition of the Food for Agile Thought newsletter, shared with 35,528 peers. This week, Packy McCormick and Markie Wagner call token maxing wasteful: AI should compile processes into code, not burn tokens at runtime; ROT (return on tokens) is essential. Deb Liu warns that chasing efficiency gains only builds a faster treadmill, while Elena Verna insists companies need employees with agency, not more agents. Roman Pichler centers emotional intelligence as the capability AI cannot replicate, Jenny Wanger swaps team health scorecards for structured conversations, and Grant Harvey examines who controls Anthropic’s Claude Fable 5.
Next, Gary Marcus questions whether AI IPOs resemble early Amazon or history’s largest capital misallocation. At the same time, Arvind Narayanan and Sayash Kapoor argue that AI only compresses execution, not decision-making, leaving engineers irreplaceable. Gaurav Savla offers PMs a practical playbook for shipping AI features, from latency budgets to drift monitoring, and Rich Mironov warns that funding software as one-time projects kills products past v1.0. Also, Sean Goedecke examines why trust between engineers and PMs erodes so quickly.
Lastly, Ara Kharazian reports that top firms spend $7,449 per employee per month on AI, with Anthropic overtaking OpenAI. Yet, Kristin Broughton, Mark Maurer, and Jennifer Williams find that only 26% of companies fully track those costs. Ruben Dominguez believes most organizations overestimate their AI maturity by two levels, and Cris Beswick adds that declining empathy and psychological safety quietly dismantle the capacity to innovate. Finally, Ben Maraney shows what structured adoption looks like through Forter’s agent sprint.
TL;DR: Compounding Systems and Agents Go Hand-in-Hand
Every AI conversation starts from zero because the model forgets who you are. The Claude Cowork Online Course teaches you to change that: build persistent Skills, connect your tools, and assemble agents for recurring work. No coding.
Thesis: “A prompt disappears after one use; a Skill compounds across every session.”
TL; DR: AI’s 1997 Internet Moment — Food for Agile Thought #547
Welcome to the 547th edition of the Food for Agile Thought newsletter, shared with 35,532 peers. This week, Benedict Evans tells Lenny Rachitsky that today marks AI’s 1997 Internet Moment and asks whether automation kills tasks or jobs, which Casey Newton’s guest, Kathryn Anne Edwards, treats as real but manageable, faulting unemployment insurance rather than fearing an idle underclass. Itamar Gilad warns that cheap AI coding tempts teams to build before validating, while Malcolm Spittler and Dylan Patel name the value GDP misses ‘Dark Output,’ and Joost Minnaar prefers autonomous-team networks over a single chain of command. Also, Marina Favaro and Jack Clark sketch the implications of recursive self-improvement of AI.
Next, Rich Mironov warns that AI ships 100x more code, yet attention and budgets don’t scale, so products that skip discovery rarely stick. METR, with Anthropic, Google, Meta, and OpenAI, finds that AI agents could plausibly go rogue but not robustly, while Teresa Torres notes that Cowork’s VM hosts the Mini Shai-Hulud worm without blocking it. Mike Fisher likens siloed teams to hand-copying the Diamond Sutra, and Jeff Gothelf redefines “done” as acceptable variance.
Lastly, Marc Abraham borrows venture capital’s ‘terminal value’ to help product managers judge whether a product merits more investment. Addy Osmani calls the opposite reflex ‘cognitive surrender,’ in which you stop thinking and accept the AI’s answer without checking it. Mark Graban shows confession works only when fixes follow, citing Burger King and Domino’s, while Tobi Lütke runs Shopify’s agent River in public Slack so everyone learns by watching. Finally, Anthropic open-sourced 11 role-specific Claude Cowork plugins for knowledge workers.
