PeachCat-Hello

What I'm Up To Now, Besides No Good

I'm a software engineer turned Claude enthusiast and founder, currently based in San Francisco. I left Tumblr in mid-2025 after a long, wonderful journey there, because we're at a very unusual, exciting time in history, and I think this decade matters more than most. There's a question I haven't been able to stop thinking about: how do we make AI go well for the many, not just the few?

I'm especially drawn to the economic side of this, what I've been calling post-labor economics. I genuinely think we have a shot at a post-scarcity future, where humans can not just survive but expand their autonomy and capacity to do things — but history doesn't happen on its own. It takes actual effort, experimentation, civic participation, a lot of people willing to not just build but *care* about what we should build. That's the work I'm most excited about right now, so I created postlabor.dev to get technologists, researchers, and builders of all stripes working on this together.

The rest of the time, I'm hacking on weird little projects (I'm a chronic hackathoner — met my husband at one!), playing with Buuta my mischievous molepig (aka OpenClaw bot), and scheming up ways to get fun illegible people in a room together. I've always cared a lot about community and scenius — I've helped build group houses in NYC and SF, hung out in philosophy salons like Interintellect, and I just think the best things happen when you get great people in proximity.

If you're in SF and care about building a better future, come say hi!

Main Projects On-Deck (as of early 2026)

A new residency in SF — I'm exploring a new residency in SF for scientists, founders, and serious thinkers who care deeply about how to make AI go well. You can see an early preview at Epilogue House. More on this soon 👀

postlabor.dev — A community for technologists, researchers, and builders who care about shaping the future of work in the AI age. We host events, run hackathons, build experimental projects like autonomous companies, and other content. I recently published a thesis laying out a 'state of the union' for the current debates around automation and the ideas that are emerging, and I maintain an auto-curated reading list of the best reads on the topic.

AIMibot — The AI tamagotchi, for us grown-up kids :) A little voice wearable that acts like a cute pet, grows as you talk with it, and comes in six different tribes.

Ad Infinitum — Won the self-improving agents hackathon. An agent that autonomously creates ad campaigns, runs them on Meta, and uses learnings from each round to improve. We're spinning this out into a kind of "reverse Kickstarter", where AI proposes and evolves campaigns using a genetic algo so humans can choose to ship the winners. Our thesis is that AI makes it easier than ever to build, but hard to know what.

🛠️ Other Projects

Naggy AI - web demo — A Flutter app I built over the holidays to help my ADHD husband actually do his tasks. It's a cute little voice agent that calls you until you do your task, and will body-double with you too like an ADHD coach. My husband says it's really helped him (and is especially effective at 1) getting him out of social events and 2) making him look incredibly busy). Most importantly, it's named after our cat Aggy, an orange with a remarkable sense of time (namely mealtime).

Homeship Pilot — An experimental project I built for me and my husband. These are personal agents that have all our context, that then talk to each other or a shared agent to coordinate tasks. The idea is to see if we can find win-win scenarios by sharing our data — like a household copilot.

MyGeneHealth — Upload your raw 23andMe data and get detailed analysis of thousands of your genetic markers, researched across multiple science sources refreshed monthly.

Upwork Mission Control — Hackathon project for the Remote Labor hackathon, where we tried to defeat the remotelabor.ai bench. This scrapes Upwork and other tasker sites for easily automatable tasks (e.g. logo design) and solves hundreds at once. The core add-on from the remotelabor bench is that an implementer works with an evaluator agent to pick the best output or retry until passing, since very few can one-shot a real task.

Detecting MechaHitler — Won the def/acc hackathon. Based on the emergent misalignment paper: can we detect unintended behavioral changes from an RL training run? Like, you train a model to be more helpful in customer service and accidentally get a gaslighting enthusiast... or you train it to be more uwu and end up with MechaHitler.

text2viz.dev — Quick AI visual diagram creator for any essay, in various styles.

kyo.bio — Kyo is a little bird who introduces you to people you want to meet. I built this for the Democratize Intelligence conference. Kyo chats with people over email to understand who they are and what they need, then makes the best mutual matches using a custom reranker under the hood. It makes a spiffy graph!

Framework Zero — An infrastructure layer to integrate human preferences into autonomous organizations. We're heading towards a fully automated future, and we need it to be one where human preferences at scale can be represented. I've been particularly interested in the digital twins piece — agents that know your preferences and can represent you. We finished the Dagihouse accelerator and have been running an event series on post-labor economics.

Autonomous Fanfic Gen — Like webtoons but for fanfic. Plug in your fandom, plot idea, tags and ships and this agent will scaffold outline, characters, scene beats before generating chapter by chapter. An evaluator loop then scores and re-gens for the best writing.

Two Degrees — A platform for matchmakers to better matchmake their friends. People are sick of dating apps, and for good reason! Intros from friends and people in your network are a much better way to meet — it just takes a better tool to do so :)

FicWorld — Inspired by the Bookworld paper, this is a story-making experiment in emergent fiction. Every character is their own agent with memory, personality, backstory, and emotional state, responding to a world driven by a World Agent and woven into stories by Plot Agents and Narrators.

⚡ Some other little things I've built in the last few months

Some of my ongoing non-coding goals:

You can follow along a running log of my random draft-tweet thoughts here. I also publish a sporadically-updated substack called everythingireadtoday that rounds up fun longreads I've found around the web - these are usually in economics, sociology, tech news, science, psychology, progress studies et al, but really just about anything that I find interesting.

Books I'm reading as of 2025:

Current/finished non-fiction reads:

Current/finished fiction reads: