The convening body for the agent era

Common standards for the agent era.

LLM Council convenes the people building large language models and agent systems to align on the standards the next internet needs — how agents remember, prove who they are, interoperate, and stay accountable.

Standards built to hold across every model — GPT · Claude · Gemini · Llama · Mistral · Grok · DeepSeek · Qwen · Kimi · GLM — open & closed.
Why now

Every lab is solving the same problems alone.

Agent memory. Identity. Verification. Interoperability. The hard questions of the agentic web are being answered in isolation, a dozen times over, in incompatible ways. The last wave of AI competed on raw capability. The next one is decided by trust you can check — between humans, between agents, and between the companies building them. That needs a table. This is the table.

Working groups

Where we align.

Memory

Portable, governed context — so agents remember and stay accountable across systems.

Identity

A shared way for agents (and the people behind them) to prove who they are.

Verification

Claims an agent makes should be checkable — provenance and signed, verifiable work.

Interoperability

Common protocols so agents from different labs can work together without lock-in.

Principles

What we hold to.

Membership

A seat at the table.

Founding members forming

LLM Council is convening its founding members by invitation — leaders and builders at the labs shaping large language models and agent systems. If that's you, request a seat.

We read every request. No spam, ever.