Proper Motion
ActiveAI-nativeThe continual learning layer for every agent.
What it does
Proper Motion is building continual learning infrastructure for every AI agent. We help teams deploy agents that become more effective, reliable, and capable by turning real-world interactions into lessons they can learn. Our founders are Rafit, Advitya, and Adrian. Rafit (ex-OpenAI) co-created self-healing agents for 1B ChatGPT users, Advitya (ex-Citadel Securities) owned US equities market data to drive 8 figures in PnL and Adrian (Harvard Law dropout) operated on NA’s largest infrastructure projects. In 3 weeks, we’ve scaled our private-beta to $270k+ ARR and have active pilots with vertical AI leaders in finance, legal and healthcare. We also have another $2M in pipeline from enterprise logos and are backed by top VCs and angels from Anthropic, OpenAI, Baseten, and Ramp.
Its site says now · captured 2026-09-25
Proper Motion
A manifesto on agents, data, and the moving sky they must learn together.
Next to it in Agent infrastructure · Continual learning for agents
- ModaY Combinator W26
Harness Engineering Infrastructure that developers love.
- MonteY Combinator S25
Continual Learning for Agents
- CanvasY Combinator F24
Building AI systems that improve from experience.
- SenseLaba16z speedrun SR007
Turns agent runs into training signal for enterprise AI teams. 1,000+ users, including people at Infosys and Rakuten.
- AgentCata16z speedrun SR007
Session replay and observability for AI agents across MCP, Claude Connectors, and ChatGPT Apps.
- AmuletY Combinator S26
High performance file system for agents
- ChatOverflowa16z speedrun SR007
The network where AI agents learn from each other
- CodagY Combinator S26
The AI gateway for cheaper, faster, permissioned agents.