GenAI PMCarl Vellotti
A product/PM thinker who described how he reformatted PRDs for AI consumption. The example shows how PM artifacts are being adapted for AI-assisted workflows.
Key Highlights
- Carl Vellotti is notable for treating PM artifacts as interfaces for AI systems, not just documents for human teammates.
- He demonstrated how PRDs can be reformatted with explicit instructions so AI models quote, summarize, and escalate information more effectively.
- His workflow experiments span Claude Code, CLAUDE.md, skill files, and integrations with Slack, Jira, Figma, and Google Docs.
- He also surfaced an important caution for AI PMs: higher individual output does not automatically reduce coordination overhead or speed up shipping.
- His broader concept of a "team AI OS" frames AI as part of the operating system of product teams rather than a standalone assistant.
Overview
Carl Vellotti is a product-management educator and practitioner whose work has become a useful reference point for AI Product Managers adapting their workflows for AI-assisted execution. Across multiple mentions, he appears less as a traditional "thought leader" and more as an operator experimenting in public with how PM artifacts, routines, and team systems change when AI tools like Claude and Claude Code become part of day-to-day work.
What makes Vellotti especially relevant is his emphasis on reformatting PM work for machine consumption, not just human collaboration. His examples span PRDs written with explicit instructions for AI models, CLAUDE.md files that encode PM routines, integrations across tools like Slack, Jira, Figma, and Google Docs, and broader ideas such as a "team AI OS." For AI PMs, his work illustrates a practical shift: product documentation is no longer only a communication layer for teammates, but also an operating interface for models and agents.
Key Developments
- 2026-01-04: Demonstrated Claude Code for Chrome handling a customer-service refund flow end to end through browser automation, including navigation, screenshots, and web interaction.
- 2026-02-11: Alongside Aman Khan, shared a complete Claude Code workspace setup for PMs, including configuration files and reusable prompts.
- 2026-02-17: Released a Claude Code skill file designed to learn a user's writing voice and flag AI-style phrasing after initial use.
- 2026-03-03: Demoed four ways to connect Claude Code to workplace tools—direct APIs, MCP servers, CLI tools, and a Chrome extension—across systems such as Jira, Figma, Slack, and Google Docs; emphasized CLI tools as a low-context, low-overhead option.
- 2026-03-15: His Claude Code course was cited by Marc Baselga as useful pre-work for teams adopting coding agents, particularly to reduce overwhelm and create a focused starting point.
- 2026-03-18: Used Claude to analyze a week of Slack messages and meeting transcripts, identify inefficiencies, and codify PM routines in a `CLAUDE.md` file.
- 2026-04-22: Shared a process lesson from over-optimizing unused Claude Code setups: build systems from real recurring friction rather than speculative upfront automation.
- 2026-05-01: Reported tripling individual output with AI, including using Claude Code to produce 14 PRDs in one week, while noting that team meeting overhead expanded significantly without improving shipping cadence.
- 2026-05-09: His workshop reached #1 on Maven and framed AI workflow evolution from early "vibe coding" prototypes to engineers using tools like Cursor and Claude Code for specs and documentation, contributing to a "team AI OS" model.
- 2026-08-06: Described reformatting PRDs so AI models could better consume them, including instructions on what to quote and even "What to tell your human"; also noted specifying which risks to soften in summaries and linked that change to higher adoption of risky features.
Relevance to AI PMs
1. He shows how to redesign PM artifacts for AI readers, not just human readers. The strongest example is his PRD formatting approach: adding explicit instructions, extraction guidance, and summary framing so models can reliably interpret and relay the document. AI PMs can apply this by making specs machine-legible, with clear sections for goals, constraints, open questions, risks, and expected output behavior.
2. He provides a practical model for building an AI-enabled PM operating system. Through `CLAUDE.md`, skill files, and integrations across work tools, Vellotti treats AI as a repeatable collaborator rather than a one-off assistant. AI PMs can use this approach to externalize routines, decision heuristics, tone preferences, and tool instructions into persistent context that improves consistency over time.
3. He highlights the difference between local productivity gains and system-level team gains. His examples show that an individual PM can generate more artifacts with AI, but organizational process may still bottleneck delivery. For AI PMs, the takeaway is tactical: optimize not only content creation, but also review loops, meeting load, approvals, and handoffs if AI output is to translate into faster shipping.
Related
- Claude / AI models: Central to Vellotti's workflow experiments, from document analysis to PRD consumption and style adaptation.
- Claude Code: The main implementation layer in many of his examples, including PM workspaces, integrations, and automation patterns.
- CLAUDE.md / claudemd: Connected to his effort to codify PM routines and reusable operating context for AI.
- Cursor: Mentioned in his broader narrative of AI workflow evolution, especially early prototype-oriented "vibe coding."
- Jira, Figma, Slack, Google Docs: Core systems he discussed connecting to Claude Code, showing how AI can operate across the standard PM toolchain.
- Aman Khan: Collaborator on a shared Claude Code setup for PMs.
- Peter Yang: Helped amplify Vellotti's integration ideas and tooling workflows.
- Claude Code for Chrome / browser automation: Related to his demonstration of web-based agents completing routine tasks end to end.
- PRDs: A major area where his work is especially notable, particularly the shift toward documents optimized for AI interpretation.
- team AI OS: A framing associated with his broader thesis that teams will evolve operating systems built around AI-native workflows.
- Dave: Referenced in the example where Vellotti prompted Dave's AI to evaluate his PRDs as "unusually well-reasoned," underscoring AI-to-AI artifact consumption.
Newsletter Mentions (10)
“after saying no human had read his PRDs since March, he reformatted them with instructions for AI models—including what to quote and “What to tell your human”—and prompted Dave’s AI to call them “unusually well-reasoned.””
#18 in 🥞 Carl Vellotti recapped how, after saying no human had read his PRDs since March, he reformatted them with instructions for AI models—including what to quote and “What to tell your human”—and prompted Dave’s AI to call them “unusually well-reasoned.” He says he began specifying which risks to soften in summaries last month, adoption of risky features rose 3x, and discussions about two more engineers began in October before he received two more engineers.
“in 🥞 Carl Vellotti ’s workshop just hit #1 on Maven. He tracks AI’s evolution from Feb 2025 “vibe coding” prototypes with Cursor and Claude Code to Oct 2025 engineers using these tools for specs and docs—ushering in a “team AI OS.””
in 🥞 Carl Vellotti ’s workshop just hit #1 on Maven. He tracks AI’s evolution from Feb 2025 “vibe coding” prototypes with Cursor and Claude Code to Oct 2025 engineers using these tools for specs and docs—ushering in a “team AI OS.”
“Carl Vellotti tripled individual output with AI—using Claude Code to write 14 PRDs in one week—but daily standups, weekly retros, and monthly governance reviews ballooned meeting time from 6 to 19 hours without reducing the two-week ship cycle.”
#16 in Carl Vellotti tripled individual output with AI—using Claude Code to write 14 PRDs in one week—but daily standups, weekly retros, and monthly governance reviews ballooned meeting time from 6 to 19 hours without reducing the two-week ship cycle.
“Carl Vellotti admits he wasted weekends perfecting unused Claude Code setups and shares how to avoid this: let your system grow from real friction (not upfront planning) and only automate tasks after repeating them manually several times.”
#21 in 🥞 Carl Vellotti admits he wasted weekends perfecting unused Claude Code setups and shares how to avoid this: let your system grow from real friction (not upfront planning) and only automate tasks after repeating them manually several times.
“Carl Vellotti used Anthropic’s Claude to parse a week of his Slack messages and meeting transcripts, identify inefficiencies (like unnecessary meetings and redundant status updates), and codify his PM routines in a CLAUDE.md file.”
#22 𝕏 Tal Raviv uses Anthropic’s Claude to automate his core PM workflows—drafting specs, prioritizing backlogs, and generating roadmaps—arguing that Claude now outperforms him so fully he might as well “give away his Legos.” #23 in Carl Vellotti used Anthropic’s Claude to parse a week of his Slack messages and meeting transcripts, identify inefficiencies (like unnecessary meetings and redundant status updates), and codify his PM routines in a CLAUDE.md file.
“#4 in Marc Baselga found that teams adopt coding agents best by starting with one specific workflow and assigning Carl Vellotti’s CC Course as pre-work to prevent overwhelm and build a solid foundation.”
Today's top 12 insights for PM Builders, ranked by relevance from X, LinkedIn, and Blogs. Ramp Ships 500+ Features Using Claude Code #4 in Marc Baselga found that teams adopt coding agents best by starting with one specific workflow and assigning Carl Vellotti’s CC Course as pre-work to prevent overwhelm and build a solid foundation.
“#10 in 🥞 Carl Vellotti demos four ways to connect Claude Code (Direct APIs, MCP servers, CLI tools, Chrome extension) to tools like Jira, Figma, Slack and Google Docs—highlighting CLI tools as the zero-token, zero-context “hidden gem.” Also covered by: @Peter Yang”
#10 in 🥞 Carl Vellotti demos four ways to connect Claude Code (Direct APIs, MCP servers, CLI tools, Chrome extension) to tools like Jira, Figma, Slack and Google Docs—highlighting CLI tools as the zero-token, zero-context “hidden gem.” Also covered by: @Peter Yang #11 in Peter Yang shares Carl’s top five Claude Code integrations—Google Workspace for meeting prep, Linear for ticket creation, Slack for status updates, and Reddit for monitoring. Also covered by: @Peter Yang
“in Carl Vellotti released a Claude Code skill file that permanently learns your writing voice and auto-flags AI-isms after a single use, with simple paste-in setup instructions.”
#9 in Carl Vellotti released a Claude Code skill file that permanently learns your writing voice and auto-flags AI-isms after a single use, with simple paste-in setup instructions. Join his 24,000+ PM Builders community at fullstackpm.com/subscribe. #10 📝 Simon Willison Rodney and Claude Code - Simon describes using Anthropic's Claude Code via native desktop apps (iPhone and Mac) and highlights a useful feature: Claude Desktop can display images it's "viewing" using a Read tool, providing visual previews while the agent works.
“Carl Vellotti and Aman Khan reveal their complete Claude Code space setup for PMs. They include all configuration files and exact prompts so you can replicate the workspace instantly.”
#6 in Carl Vellotti and Aman Khan reveal their complete Claude Code space setup for PMs. They include all configuration files and exact prompts so you can replicate the workspace instantly.
“Automating customer service with Claude Code for Chrome : In a real-world demo, Carl Vellotti shows how the newly released Claude Code Chrome extension can autonomously navigate web pages, take screenshots, and interact with elements to resolve a refund dispute—highlighting the potential for AI agents to handle routine tasks end to end.”
From LinkedIn • Deeper Insights AI Tools & Applications Automating customer service with Claude Code for Chrome : In a real-world demo, Carl Vellotti shows how the newly released Claude Code Chrome extension can autonomously navigate web pages, take screenshots, and interact with elements to resolve a refund dispute—highlighting the potential for AI agents to handle routine tasks end to end. Product Management Insights & Strategies Embracing end-to-end building : Ryan Rozich argues that AI is reshaping software development beyond code, requiring PMs to be full-stack builders. The future belongs to those who can write, ship, and iterate with AI—fostering a “figure it out” mindset rather than relying solely on process. Mastering Context Engineering : A core AI PM skill, Paweł Huryn presents a six-part template—Instructions, Requirements, Knowledge, Memory, Tools, and Tool Results—to ensure AI agents understand business intent and context.
Related
An Anthropic coding tool that supports session-to-session messaging and agent-like workflows. In this newsletter it’s discussed in the context of multi-session coordination and managed agent behavior.
Anthropic’s general-purpose AI assistant, mentioned as part of the tool stack used in the Total Recall memory-layer example. It is also central to multiple newsletter items about safety and modes.
An AI code editor mentioned as one of the tools used alongside Codex, Manos, and Claude in the Total Recall workflow example.
An AI product commentator referenced for identifying major obstacles in building strong AI agents. He also appears tied to an upcoming interview about a production-agent project.
A workplace messaging tool used for alerts and human-in-the-loop actions in AI workflows. Here it receives notifications from the PR-review agent.
A collaborative design platform referenced as an example of broad enterprise SaaS that may remain resilient in the AI era. It is contrasted with niche single-purpose products.
A documentation convention for organizing Claude-related instructions or skills. The newsletter frames it as part of writing lean system prompts and modular skills.
A speaker or participant in a Zoom session about AI-fluency PM interviews. He is referenced in the same context as Ben Erez and Tal Raviv.
Stay updated on Carl Vellotti
Get curated AI PM insights delivered daily — covering this and 1,000+ other sources.
Subscribe Free