automatedbureaucracy.com - Automating AI Safety Research with Multi-Agent Language Simulations

This video is the first prototype capability demonstration for the multi-agent framework I am building with my startup team of young developers at automatedbureaucracy.com

The goal is to build an ethics-to-code pipeline, and build the systems I have imagined through my narratives in code. I have been recruiting a small team of young developers across India and the U.S. We plan to pitch to YCombinator in late February and are currently seeking investors.

The framework uses Langchain to simulate a collective intelligence and logs the simulation artifacts using a local ML Flow server. The codebase is written in Python, uses an object oriented framework with self-documenting code, extensive logging, and is entirely generated by models in the OpenAI family. The prototype is based on the open source project I released in my video on running local agentic AI models for personal databases. You can watch that here:
https://youtu.be/yJHytmJ4n9Q

I was inspired to found this startup by the intelligence that emerges in my science fiction novel "Simulated Singularity", which also inspired this channel. You can watch that here:
https://youtu.be/TRuz9uRlqnE

If you're interested in partnering with my team or contracting us to build custom multi-agent automation solutions, reach out to
blakedehaas@automatedbureaucracy.com
and we can collaborate.