David Markley

After spending time with clients and partners, the constraint is clear: the system is starved of high-quality collateral. Meanwhile, perpetuals are becoming the core financial layer, concentrating liquidity and streamlining access to leverage. Unity is built to sit at that intersection🐕

@bcherny How does credential sharing work? Can you create ephemeral credentials, à la 1Password, for Claude?

We're launching Claude Tag today. Tag Claude into Slack and it works in channel with you. It’s proactive, multiplayer, with its own identity and memory. But it’s not just a bot in Slack. Over the last few months, it’s totally changed how we use Claude

@karpathy @noahzweben @karpathy - for Claude, does this act as a functional replacement* of Centaur, by Paradigm? *Assuming you’re comfortable trusting your full stack to Anthropic paradigm.xyz/2026/05/open-s…

This is a new paradigm for interacting with Claude that is significantly more "inline" with all the other human activity org-wide. Once you do all of the under the hood engineering work to make this "just work" (e.g. across tools, integrations, compute environments, memory, security, etc.), Claude basically joins the team in a seamless way - you can talk to it as you would talk to a person and it can help with a very large variety of workloads. Imo this is the 3rd major redesign of LLM UIUX. The first paradigm was that the LLM is a website you go to, the second was that it is an app you download to your computer. This third one is that it is a self-contained, persistent, asynchronous entity with org-wide tools and context, working alongside teams of humans. It really takes a while to wrap your head around it, but it works and it is awesome.

@kunchenguid I like the “captain” as a drift tracer; that’s brilliant. might start implementing across all sessions. I’ll take firstmate for a spin, thank you!

i currently often juggle between about 5 projects and have been running smoothly i have a trick - firstmate has system prompt instructing it to always address me as “captain” in every response. so whenever i notice it forgot to do that, i know it’s drifting because context is noisy and i would compact or start fresh and i did put in some efforts to make sure firstmate curates its context and a lot of things get persisted instead of relying on the context window

time to reveal the next big gun in my agentic engineering setup - Firstmate! it's the only agent session i directly talk to now. if you find it mentally exhausting to juggle between all the agent sessions, this is the solution free & open source as always - details below 👇

@jeremykauffman @L0m3z @ChHouseBooks Did you say NH homeschool groups? 👀

@L0m3z @ChHouseBooks This looks great. Will be ordering a set and likely promoting to New Hampshire homeschool groups

Today our partners @ChHouseBooks have launched an amazing new project. For any child K-6 you can buy a full year of expertly curated reading in one easy click. It includes box sets of remastered classics that introduces young readers to Homer, Shakespeare, Beowulf and the foundations of the Western Canon. As well as carefully curated book bundles that cover everything from nature study to handwriting, beginner's Latin to math, to the best of school-age literature from Tolkien to Milne. One stop shop. No fuss. Each box sets also comes with a 70-80 pg teaching guide that helps parents best deliver this material to their children. Perfect for homeschoolers or any parent who just wants more for their kids. Available for pre-order now. chapter.house

h/t @garrytan for "fat skills, thin harness" & @claudeai for cooking on Claude Managed Agents

@claudeai @bcherny - if a solopreneur was tinkering via their personal account and would like access to Outcomes, Multiagent Orchestration & Memory, is there a path? Or is the answer to spin up a business account, migrate to working there, and then apply? Thanks in advance!

Build and deploy your agents through the Claude Console, Claude Code, or our new CLI: platform.claude.com/workspaces/def… Read more on the blog: claude.com/blog/claude-ma…

Introducing Claude Managed Agents: everything you need to build and deploy agents at scale. It pairs an agent harness tuned for performance with production infrastructure, so you can go from prototype to launch in days. Now in public beta on the Claude Platform.

@trq212 yes x.com/dbmarkley/stat…

I want to do some streams where I work with non-technical people using Claude Code to figure out how they might be able to improve their process. My feeling is that just a few tips could make a big difference in efficiency. Any mutuals interested?

Seeing a lot of builders post today that they were also building agent harnesses and orchestrators, thinking they were one of maybe 50 people with this idea. Turns out there were thousands of us. Which raises an interesting question: Did Claude lead us toward patterns that Anthropic had already discovered internally? Or did Anthropic watch how we were all building with Claude and productize what emerged? Either way, it's a fascinating feedback loop. The tool shaped how we built. How we built shaped the next version of the tool

@nikil thank you for the inspiration @nikil 🫡

@dbmarkley 🙌

💯 The stack I built is specific to how I operate as a PM running multiple projects. The orchestrator, knowledge architecture, and review pipeline all emerged from my specific constraints. Anthropic's Managed Agents will handle the generic infrastructure better than I ever could. The question I'm most interested in is: what are the domain-specific patterns that sit on top of that infrastructure? The committee design, the verification fidelity spectrum, the knowledge routing. Unclear how we share those templates at scale as I don't think they get commoditized by Anthropic/OpenAI

@dbmarkley We're just moving up the stack here dude. I think the riches are in the niches. If you're broad, Anthropic or OpenAI is going to eat your lunch. The trick is to get really specific.

@austin_way Isn’t the conclusion to use locally hosted open source models so they can’t take it away?

A student with a textbook will outperform a student with ChatGPT. Hear me out. A team at UPenn took about 1,000 high school students and split them into two groups. One group got ChatGPT to help them with their math practice. The other group got nothing but their textbook and notes. The ChatGPT group scored 48% higher on homework than the textbook group. By every metric, it looked like the question was already answered. Then the researchers took ChatGPT away and gave everyone the same exam. The ChatGPT group scored 17% worse than the students who never had it. That is insane. None of this means AI can't work in education. It means a raw chatbot is not a learning tool. (Peer-reviewed, published in PNAS, 2025.)

@fancylancer3991 My intuition is that the generalization works well across verifiable tasks, thus the work is to transform subjective tasks into objective approximations in order to benefit from the generalized intelligence

It's not new that agents can be self-improving. The most interesting thing for me is that HyperAgents improvements generalize beyond one domain. They tested it on coding, paper review, robotics rewards functions, and Olympian-grade math. The agent was able to beat previous self-improving mechanism across all of them. Which means, in theory, if you have a very well working agent that can self-improve you can just keep working with this agent on everything you want and it'll get better at any other future you give it. So for example you work with your agent on coding and research. Then you switch to marketing and the same agent can learn faster and better then any marketing-specific agents. Just theory, but a cool way to think about.

@nikil @davehappyminion Install ClawFlows... ... immediately feel like you live in the future

Today I'm open sourcing my secret project I use 1000+ times daily Introducing ClawFlows - workflow system for OpenClaw - simple, reliable, powerful - instantly enable 100+ prebuilt workflows @davehappyminion runs my life with ClawFlows: - morning briefing: weather, messages, daily inspiration - meeting prep: research & brief me on who I'm meeting - life coach: reads my health data & suggests improvements and many more... Been using it daily for 1.5 months and massively leveled up my life. Dave and I poured a lot of love and energy into this to help your openclaw improve your life! enjoy ❤️

tried “remove HITL” system ran better so much for all those RTS hours

Our co-founder & COO @dbmarkley will be live on X today at 12 PM ET discussing real-world stablecoin adoption and where the next wave of usage is coming from. Tune in 👇