GenAI PMGarry Tan
A named builder/leader who used Claude-generated code to fix a Dockerfile PATH issue in OpenClaw. The mention illustrates practical AI-assisted debugging.
Key Highlights
- Garry Tan is portrayed as a hands-on AI builder using agentic tooling for real product development, debugging, and retrieval system design.
- His work on GBrain emphasizes practical AI PM concerns like eval harnesses, hybrid retrieval, memory structure, and multi-repo context.
- He advocates clear architecture diagramming and DRY context management, making complex agent systems easier to reason about and ship.
- His OpenClaw and Hermes comparisons highlight an important PM tradeoff between raw capability and operational reliability.
- The Dockerfile PATH fix using Claude-generated code is a concrete example of AI accelerating engineering troubleshooting.
Garry Tan
Overview
Garry Tan appears in the newsletter as a hands-on builder and operator working at the intersection of AI agents, developer tooling, retrieval systems, and product experimentation. Across mentions, he is associated with OpenClaw, Hermes Agent, and the open-source GBrain project, using frontier models and agent workflows not just for demos but for real product-building tasks like debugging Dockerfiles, designing architecture, evaluating retrieval quality, and turning personal data into useful AI-driven experiences.For AI Product Managers, Garry Tan matters because his examples are highly practical: use AI coding systems to unblock engineering work, diagram agent architectures clearly, build eval harnesses early, reduce redundant context via source control, and iterate in public on cost, speed, and usability tradeoffs. His mentions consistently illustrate an applied AI PM playbook centered on shipping, instrumentation, and learning from real workflows instead of abstract AI strategy alone.
Key Developments
- 2026-04-18: Launched GBrain, an open-source AI assistant designed to be built directly into OpenClaw or Hermes Agent workflows.
- 2026-04-20: Released GBrain v0.13, improving graph queries by having OpenClaw/Hermes extract properties into YAML for easier downstream retrieval.
- 2026-04-23: Announced that GBrain supports multiple repos per brain, enabling storage of GStack code transcripts, plans, and Claude Code artifacts in one system.
- 2026-04-27: Built a GBrain eval harness using 145 queries over an Opus-generated corpus with a hybrid retrieval stack combining graph, vector, and grep.
- 2026-05-01: Featured in a conversation with Demis Hassabis about how Google DeepMind turns research breakthroughs like AlphaGo Zero and AlphaFold into products while scaling safely toward AGI.
- 2026-05-02: Imported 17 years of Foursquare check-in data into an OpenClaw/Hermes workflow to generate personalized travel guides; also released GBrain v0.25 for benchmarking AI evaluations against real-world brain queries.
- 2026-05-03: Compared Hermes Agent to a reliable “Honda Accord” and OpenClaw to a powerful but finicky “Ferrari,” highlighting the product tradeoff between stability and performance.
- 2026-05-04: Suggested that GBrain should use git history to fetch context on demand, reducing repeated inputs and reinforcing the DRY principle.
- 2026-05-10: Recommended diagramming AI agent codebases and architecture in plain ASCII and interrogating each component to improve design clarity and development speed.
- 2026-05-10: Shared that he spends about $2K/month on OpenClaw AI tokens, describing a “tokenmaxxing” approach to accelerate development and startup insight generation while aiming to make these capabilities broadly affordable.
- 2026-05-11: Used Claude-generated code to fix a PATH misconfiguration in OpenClaw’s Dockerfiles, a concrete example of AI-assisted debugging that quickly restored development flow.
Relevance to AI PMs
1. Treat AI as an execution layer, not just a feature. Garry Tan’s Dockerfile debugging and architecture work show how PMs can use coding agents and LLMs to shorten iteration cycles, unblock teams, and validate technical assumptions faster.2. Build evals and retrieval quality checks early. His GBrain eval harness, hybrid retrieval design, and benchmarking work underscore that AI products need measurable evaluation frameworks, especially when retrieval, memory, and agent behavior are core to product value.
3. Design for context efficiency and system clarity. The emphasis on ASCII architecture diagrams, git-history-based context retrieval, and DRY workflows points to a tactical PM lesson: better context management lowers cost, improves agent reliability, and makes multi-component AI systems easier to ship and maintain.
Related
- OpenClaw: A major part of Garry Tan’s workflow; mentioned in connection with debugging, token usage, and high-performance agent development.
- Hermes / Hermes Agent: Frequently paired with OpenClaw and GBrain; positioned as a more stable counterpart in his product comparisons.
- GBrain: His open-source AI assistant project, central to many mentions involving memory, retrieval, evals, and multi-repo support.
- Claude / claude-code / Anthropic: Connected through Claude-generated code and Claude Code artifacts, illustrating practical AI-assisted development.
- Opus: Used in generating a corpus for GBrain evaluation, tying Tan’s work to model-driven benchmark creation.
- Demis Hassabis / Google DeepMind: Linked through a discussion on translating AI research into products and safely scaling advanced systems.
- Architecture / DRY / tokenmaxxing: Recurring themes in his mentions that connect to product design clarity, context efficiency, and AI cost-performance tradeoffs.
- gstack, gbrain-repo, gbrain-v013, gbrain-v025: Related implementation artifacts and versions that show fast iteration on the broader GBrain ecosystem.
Newsletter Mentions (14)
“Garry Tan spent the morning diving into OpenClaw’s Dockerfiles to fix a PATH misconfiguration using Claude-generated code.”
#3 𝕏 Garry Tan spent the morning diving into OpenClaw’s Dockerfiles to fix a PATH misconfiguration using Claude-generated code. By afternoon the bug was squashed and development was back on track.
“#4 𝕏 Garry Tan suggests diagramming your AI agent codebases and architecture in plain ASCII, then relentlessly questioning each component to clarify design and accelerate product development. #9 𝕏 Garry Tan spends $2K/mo on Openclaw AI tokens to turbocharge product development and startup insights. He’s “tokenmaxxing” now with a goal to make these capabilities affordable for everyone in 18 months.”
GenAI PM Daily May 10, 2026 GenAI PM Daily 🎧 Listen to this brief 3 min listen Today's top 11 insights for PM Builders, ranked by relevance from X, Blogs, and LinkedIn. PromptLayer’s multi-step agent evaluation framework #1 𝕏 Jason Zhou launched `/goal` support in CodeX and Hermes agents for one-step autonomous coding, advising use of interview mode, clear stop conditions, and a goal-buddy to manage state and goal files. #2 📝 PromptLayer Blog What Is Agent Evaluation? A Practical Guide for AI Teams - Agent evaluation tests whether an AI agent reliably completes tasks across real inputs, edge cases, and new versions by scoring not just final outputs but multi-step behavior via black-box, trajectory, and component-level evaluations, using metrics like task completion rate, tool selection accuracy, unsupported-claim rate, latency/cost per step, and regression pass rate. PromptLayer offers tracing with span-level context, reusable datasets, batch evaluations, backtesting, regression testing, automated evaluation triggers on new prompt versions, and flexible pipelines including code execution, human input, conversation simulation, regex checks, and LLM assertions. #3 in Udi Menkes built his new product’s entire data flow in a single interactive HTML file—complete with diagrams, in-page navigation, and color-coded complexity—letting his team understand it in minutes instead of hours. #4 𝕏 Garry Tan suggests diagramming your AI agent codebases and architecture in plain ASCII, then relentlessly questioning each component to clarify design and accelerate product development. #5 𝕏 Boris Cherny says Claude Code’s switch to a native installer means npm-only stats undercount its real usage. On Thursday it hit its second-highest signup day ever with 15× growth since Jan 1—now you can ask Claude to debug your SQL. #6 𝕏 Boris Cherny is enhancing Claude Code’s UX for snappier performance and adding debug logs so users can self-serve hang diagnostics. #7 𝕏 Harrison Chase calls LangSmith an org-wide platform for building AI agents that speeds up cross-functional collaboration and tightens feedback loops. #8 𝕏 Santiago showcases a step-by-step guide for constructing Python-powered multi-agent systems from scratch, leveraging MCP and A2A patterns to incrementally add complexity and enable collaborative AI agents. #9 𝕏 Garry Tan spends $2K/mo on Openclaw AI tokens to turbocharge product development and startup insights. He’s “tokenmaxxing” now with a goal to make these capabilities affordable for everyone in 18 months. #10 𝕏 Harrison Chase argues that treating AI agents as systems to measure and iteratively improve isn’t just a technical challenge—it demands intentional human collaboration and team processes. #11 in Peter Yang warns that unedited AI-generated markdown can compound small errors over time—what starts as 5% “slop” quickly balloons into an overwhelming pile of confusing, unverified content. Found this valuable? Share it with another PM - they can subscribe at genaipm.com Unsubscribe • Switch to Weekly
“𝕏 Garry Tan suggests Gbrain should leverage git history to fetch context on demand, avoiding redundant inputs and adhering to the DRY (“don’t repeat yourself”) principle.”
#11 𝕏 Garry Tan suggests Gbrain should leverage git history to fetch context on demand, avoiding redundant inputs and adhering to the DRY (“don’t repeat yourself”) principle.
“#12 𝕏 Garry Tan likens Hermes Agent to a rock-solid Honda Accord and OpenClaw to a high-performance Ferrari that demands roadside tinkering but delivers exceptional power.”
#12 𝕏 Garry Tan likens Hermes Agent to a rock-solid Honda Accord and OpenClaw to a high-performance Ferrari that demands roadside tinkering but delivers exceptional power.
“Garry Tan imported 17 years of Foursquare check-in data (5,000+ entries) into his OpenClaw/Hermes platform to auto-generate personalized travel guides, starting with his top spots in San Francisco.”
Garry Tan imported 17 years of Foursquare check-in data (5,000+ entries) into his OpenClaw/Hermes platform to auto-generate personalized travel guides, starting with his top spots in San Francisco. Garry Tan released GBrain v0.25 to let contributors benchmark AI evaluations against their own real-world brain queries.
“Garry Tan sits down with Demis Hassabis to unpack DeepMind’s playbook for turning research breakthroughs (AlphaGo Zero, AlphaFold) into real-world products and charting strategies for safely scaling toward AGI.”
#13 𝕏 Garry Tan sits down with Demis Hassabis to unpack DeepMind’s playbook for turning research breakthroughs (AlphaGo Zero, AlphaFold) into real-world products and charting strategies for safely scaling toward AGI.
“Garry Tan built a GBrain eval harness using 145 queries over an Opus‐generated corpus and a hybrid retrieval stack (graph, vector, grep).”
#1 𝕏 Garry Tan built a GBrain eval harness using 145 queries over an Opus‐generated corpus and a hybrid retrieval stack (graph, vector, grep).
“#17 𝕏 Garry Tan announced that GBrain now supports multiple repos per brain, paving the way to store your GStack code transcripts, plans, and Claude Code artifacts directly in GBrain.”
#17 𝕏 Garry Tan announced that GBrain now supports multiple repos per brain, paving the way to store your GStack code transcripts, plans, and Claude Code artifacts directly in GBrain.
“#7 𝕏 Garry Tan dropped GBrain v0.13, upgrading graph queries by letting OpenClaw/Hermes extract properties into YAML so data isn’t just ingested but processed for easy retrieval.”
#7 𝕏 Garry Tan dropped GBrain v0.13, upgrading graph queries by letting OpenClaw/Hermes extract properties into YAML so data isn’t just ingested but processed for easy retrieval.
“Garry Tan launched GBrain, an open-source AI assistant you can build directly into your OpenClaw or Hermes Agent (repo: github.com/garrytan/gbrain).”
#13 𝕏 Garry Tan launched GBrain, an open-source AI assistant you can build directly into your OpenClaw or Hermes Agent (repo: github.com/garrytan/gbrain). #14 ▶️ I tested Seedance 2.0. Wow. Greg Isenberg Cense 2’s multi-input video editor in the Enhancer platform is used to generate and edit 720p videos by combining up to two images, two videos, and one audio file via tagged natural language prompts in about 60 seconds.
Related
Anthropic’s coding-focused assistant/tool used for building and automating engineering workflows. The newsletter references it in both security and product-usage contexts.
AI company behind Claude and related developer tools. In this newsletter it is highlighted for internal use of Claude Code and for product expansion into legal workflows.
Anthropic’s assistant/model family, referenced in enterprise deployment, managed agents, and coding workflows. For AI PMs, it is central to agentic product design and enterprise integration.
Product and growth writer/podcaster focused on startups and PM topics. He is cited here for commentary on Anthropic’s operating pace and PM compensation content.
A software project/company referenced as the codebase Garry Tan worked in while fixing a Dockerfile PATH issue with AI-generated code.
Google’s frontier AI research organization. The newsletter references it for launching interactive experiments in Google AI Studio.
An AI software company behind Devin, a coding agent. Important for PMs evaluating automated bug fixing and enterprise engineering workflows.
Co-founder and CEO of Google DeepMind. He is mentioned here in relation to new funding for Isomorphic Labs and a Gemini-powered UI prototype.
A company or product referenced as a candidate for leveraging git history to fetch context on demand. The implication is a product design focused on context reuse.
A large language model used here to generate a corpus for retrieval evaluation. In AI PM contexts, it is relevant as a model choice for content generation and analysis tasks.
A personal AI agent compared in a benchmark roundup. Useful for PMs looking at alternative agent systems and workflow automation.
A communications platform used here as a runtime/connection endpoint for personal AI demos. It is mentioned alongside WebRTC in a quick setup workflow.
An agent tool described as reliable and solid in the newsletter's analogy. It is contrasted with OpenClaw as the more dependable option.
Stay updated on Garry Tan
Get curated AI PM insights delivered daily — covering this and 1,000+ other sources.
Subscribe Free