Open-source infrastructure for AI partnerships that persist.
Your AI forgets you every conversation. Your preferences, your working style, the decisions you made yesterday, the project context from last week. Every session starts cold. You repeat yourself. The AI guesses. Nobody builds on what came before.
The Intent Layer fixes this. It gives any AI client (Claude, Copilot, ChatGPT, local models) a persistent identity layer: who the human is, how they work, what's happening, and how the AI should behave. One system, any model, every conversation starts warm.
AI Command Center -- Template repo. Clone it, fill in your identity files, and your AI partnership has structure. Personalities define how the AI behaves in different contexts. Tracking areas organize your work. Status captures what's happening across everything. Works today with Claude Projects and a GitHub repo.
MCP Server -- Runtime API that serves the AI Command Center to any MCP-compatible client. Your identity, personalities, status, and work items accessible from Claude, Copilot, ChatGPT, or a local model running on your laptop. One server, every client.
theintentlayer.com -- Product home. Website, OAuth provider, and (coming soon) an interactive onboarding flow that walks you through setting up your own AI Command Center.
The system has three layers:
- Identity rules -- who you are, how you work, what rules the AI must follow. Universal across every conversation.
- Personalities -- behavioral modes the AI switches between. Ops mode thinks about systems. Book mode thinks about writing. Consulting mode thinks about the client. Each loads different context and follows different rules.
- Status -- what's happening right now. Active projects, deadlines, recent decisions. Synthesized daily so every conversation starts current.
A 14B open-weight model running locally can read these files and operate as your partner. A frontier model like Claude does it even better. The architecture works independent of model capability. That's the point.
AI Trust Commons -- The governance and research organization behind the Intent Layer. Published the HIP Charter, submitted to NIST NCCoE, and maintains the open-source governance framework for AI agents.
Nikhil Singhal -- 25 years building and leading engineering at Microsoft, AT&T, T-Mobile, Expedia Group, and Hitachi Consulting. The Intent Layer grew from building his own AI partnership system daily with Claude, then extracting the architecture so anyone can use it.