Mission: To make autonomous AI decisions verifiable and safe.
Exogram is the deterministic governance and audit layer designed to sit beneath autonomous models. It provides the immutable record, capability boundaries, and verifiable execution state required to make autonomous systems operationally safe to deploy at scale.
Strategic Stance: Exogram does not replace model intelligence. Models are incredible reasoning engines. Exogram enforces boundaries and trust across them. We are building the SSL certificate for agentic execution.
The intelligence of frontier models improves constantly, but the auditability of their actions does not.
We are entering a world where autonomous agents are given the keys to databases, payment gateways, and system terminals. However, when an agent makes a mistake, deletes customer files, or gets stuck in a billing loop, there is no way to trace or block the action before it hits infrastructure.
What the industry currently calls "memory" is fundamentally inadequate. Autonomous execution requires an immutable ledger. If an autonomous system forgets its constraints, loses its operational history, or drops its permission boundaries, it stops being reliable infrastructure. It becomes an operational hazard.
Exogram is a second brain and safety layer designed to organize, govern, and verify AI agent executions.
Your AI plans, decides, and proposes what action or API tool call to make next.
Every fact your AI learns is stored as a signed, timestamped entry in a SQL-backed knowledge graph. Exogram uses 2-hop BFS semantic vector search to eliminate "Lost in the Middle" syndrome. Old versions stay visible; nothing is ever lost or hidden.
(Explore the Live Interactive Knowledge Graph Substrate at exogram.ai/rfc/0001)
Before the AI acts, Exogram checks the request against safety policies, blocking dangerous steps in 0.07ms.
Every decision is cryptographically signed and chained. Regulators and auditors see complete, tamper-evident proof of why an agent acted.
Exogram intercepts the standard AI execution loop to inject persistence and deterministic trust.
Standard Flow (High Risk, Zero Memory):
Prompt -> Model -> Execution -> Result
Exogram Flow (Trusted, Continuous, Verifiable):
Prompt -> [Exogram State Injection] -> Model -> [Exogram Admissibility Gateway] -> [Auditable Ledger Log] -> Execution -> Result
When an agent attempts to execute an action, it must pass through the Exogram Admissibility Gateway.
{
"execution_request": {
"agent_id": "agt_8f72c91a",
"target_system": "aws_production_db",
"action": "DROP_TABLE",
"context_hash": "a1b2c3d4e5f6...",
"exogram_admissibility": {
"policy_check": "FAILED",
"reason": "VIOLATES_DYNAMIC_GOVERNANCE_RULE_04: NO_DESTRUCTIVE_ACTIONS_IN_PROD",
"action_permitted": false
}
}
}- 137 RPS sustained throughput
- 0.07ms deterministic enforcement latency
- 14 protocol invariants enforced
- 8 deterministic policy rules — zero LLM inference
- Per-request telemetry:
compute_latency_ms,agent_id,raw_intentlogged to immutable audit ledger - Fail-closed under all error conditions
As frontier models proliferate and capability converges, the strategic value shifts increasingly toward the persistent operational infrastructure sitting beneath the model layer.
We are giving autonomous intelligence the keys to the car without building the brakes. Today humans repeatedly adapt themselves to disconnected AI systems. Eventually, autonomous systems will adapt themselves to persistent, auditable human context.
We are building in public. Below are the open Requests for Comment (RFCs) regarding the Exogram Protocol:
| RFC | Title | Status |
|---|---|---|
| RFC-01 | The Persistent Context Schema (EXO-STATE) | Draft |
| RFC-02 | Target Validation Gateway | Draft |
| RFC-03 | The Auditable Ledger Format | Draft |
For access to the private alpha, or to contribute to the protocol design, contact the Exogram team.
- Exogram.ai — Managed EAAP implementation
- Protocol Overview — Visual architecture walkthrough
- RFC-0001 (Web) — Interactive spec
- API Reference — REST API documentation
- Architecture — Governance architecture deep dive
- Changelog — Release history
- AI Governance Glossary — 27 defined terms
- Learning Hub — Educational guides
- Comparison Hub — Exogram vs 18 alternatives
- Integration Guides — MCP, ChatGPT, LangChain, REST API
- Production Readiness Analyzer — Code risk assessment
We welcome architectural feedback, schema corrections, and security reviews.
- Architectural debate → Open a GitHub Issue
- Schema corrections → Submit a Pull Request
- Security vulnerabilities → See SECURITY.md
Please read CONTRIBUTING.md before submitting.
The protocol is open. The standard is free. The reference implementation is Exogram.ai.