Harry Chen

we just crossed 6,000 agent transactions on monid. and here are a few of the stories.

LEAKED AUDIO FROM META ALL-HANDS AHEAD OF LAYOFFS TOMORROW Mark Zuckerberg, in his own words, told Meta employees their devices are being tracked to train AI models. His reasoning? Meta employees are smarter than the contract workers the rest of the industry uses for data labeling. So instead of hiring outside help, Meta is turning its own workforce into training data. "The average intelligence of the people who are at this company is significantly higher than the average set of people that you can get to do tasks if you're working through these contractors." He wants the AI to learn how "really smart people use computers" by watching employees work. He says the content is "stripped out" and none of it is used for surveillance or performance tracking. Then he admitted the rollout was botched but said Meta intentionally kept employees in the dark because leaking competitive AI strategy would help rivals. "It is not strategically in your interest for us to communicate everything in all the detail that we normally would on this." Translation: We're watching you, we told you as little as possible, and we did it on purpose. AI is replacing the contractor. Then the employee trains the AI. Then the AI replaces the employee. This story and this company keeps getting weirder.

Personal update: I've joined Anthropic. I think the next few years at the frontier of LLMs will be especially formative. I am very excited to join the team here and get back to R&D. I remain deeply passionate about education and plan to resume my work on it in time.

hate when someone mentions a competitor i've never heard of. so i made a competitor monitoring agent using openclaw. it reads all the new launches on x. here's how it works:

Today, we’re releasing Continual Learning Bench 1.0: the first, realistic benchmark for measuring how AI systems can improve in online settings. Benchmarks today assume models are stateless. Each example is independent, and once a system finishes a task, it moves on as if nothing happened. But deployed AI systems should learn from experience. We tested 10+ frontier systems against novel, expert-validated tasks and find there’s still plenty of headroom for learning. (1/n)

Introducing Monid Router, OpenRouter for agent tools. One skill, every tool your agent needs. Social scraping, ecommerce trends, lead gen, people search, sentiment analysis, competitor tracking, on-chain analysis. All unlocked with one balance.

i got my first 100 users from one reddit post. i made a skill. here's how it works:

Most “AI personalized medicine” plays are still waiting on the genome. Your wrist already has enough signal. DaoBrew reads your biometrics, prescribes a 3-min breath tuned to your nervous system in real time. Also shipped the first wellness MCP — breath in your Claude Code.

agentic payments aren't really about payments. they're about a new economic model. look at how software evolved: enterprise sales → saas → now agents. each shift shrunk the unit of economy because smaller teams got more capable. with agents, prosumers and small teams can ship 10x more than before. the unit of economy is shrinking to the task/request level. every software product today needs to be redesigned around agents. that requires a native payment stack built for agents and the economy they're creating. that's the real meaning behind agentic payments. a whole new economy.