Introducing @PoeticHQ: a new AI system that executes complex multi-hour tasks with 99%+ accuracy and 10x fewer tokens than agents.
We raised $50M at $500M from Kleiner Perkins, Founders Fund, First Harmonic, and Genius Ventures to build AI that does complex work inside Fortune 500 companies without hallucination.
While code is too brittle, agents are too unpredictable. The work that runs the global economy - anti-money laundering, fraud investigations, underwriting - needs extreme accuracy.
So we built a new kind of software that pairs the flexibility of AI with the predictability of code.
When the world stays the same, Poetic runs fixed code: fast, cheap, identical every time. When the world changes, Poetic uses AI to regenerate its approach and find its way back to the objective.
In one year, we went from zero to an eight-figure run rate as a team of four.
Since then, we’ve scaled the team and executed the highest-stakes processes at AIG, SoFi, and Chime. At SoFi, a large US bank, Poetic reached 99%+ quality on fraud investigations in five weeks.
curious -- markdown for SOPs makes total sense and since they're probably being updated/used by agents frequently version control matters, which kind of makes github the obvious answer. but, do the non technical folks on your team get scared off by it? or is it just that there's nothing better than github to store them in?
another thing I’ve noticed is that you need agents to proactively file information into a brain for it to be useful. but even with a great MCP with detailed instructions & tool descriptions to proactively file information, Claude Cowork/Chat strip those out and don’t autoload tools to protect context window - which is understandable - but makes it so that tools that might do the filing can’t just proactively/ambiently work to make the system successful.
can try to solve that with a helper skill to pair with the mcp, but gets to the same problem you mention about not having enough control over employee skills. hard to make sure everyone has the same implementation.
unfortunately, we're all a bit brainrotted. myself included.
that's why I made focusfocus, a live monitor for your attention span. when you look away from your computer to go doomscroll reels, a light ambient buzz plays at increasing volume in your headphones until you come back to what you were working on
thanks @contra & @figma for the #figmamakeathon! always a fun time building with ya
check it out here: https://t.co/CZR1Hpt0WQ
Meet Your Inner Council
Stop overthinking in a vacuum. Inner Council is voice-first AI experience where you summon different archetypes to help you think through... whatever you're thinking through.
Pick who you want at the table, ask your question, and see what they say. Almost like a Magic 8-ball, but with multiple personalities.
It's not about getting *the answer*. It's about getting perspective.
Check it out and let me know if it gives ya any good advice!
Built entirely in @figma Make. Thanks to @contra & @figma for another cheeky designathon / hackathon / #figmamakeathon
AI labs and quant funds are converging.
Why was DeepSeek built by a Chinese hedge fund and not a US tech giant?
In reality, they’re structurally the same pipeline: data -> model -> constraints -> execution -> feedback.
can’t stop thinking about my husband’s great aunt telling me that she set her name in Claude as ‘Silly’
so that it greets her like “What are you thinking about today, Silly?”