THE ENGINE

ProjectCore

The API-only control plane.
Agent runtime, memory, governance, evaluation, ledger, and observability - versioned and production-ready.

This is not a UI product.
This is the infrastructure layer.

VIEW THE API → hello@project0.io
WHAT CORE PROVIDES
AGENT RUNTIME

Register agents. Define principles. Scope authority. Execute runs with full trace persistence. Every invocation is attributable. Nothing runs without a record.

MEMORY

Three tiers. One boundary. Ephemeral context fades by design. Brain is mutable working memory - supersession-based, never silent edits. Ledger is immutable, append-only, tamper-evident. Promote is the only path across.

GOVERNANCE

Rules that execute, not documents that sit. Halt, suspend, resume - all recorded. Authority modes declared at commit time, never inferred after the fact.

EVALUATION ENGINE

Open a bounded decision window. Run agents against real decisions. Close it. Seal it. The artifact is independently verifiable by anyone who asks.

LEDGER

Append-only. Tamper-evident. Authority-attributed at the moment of commitment. Not a log. A system of record.

OBSERVABILITY

Run records. Execution traces. System event log. Audit export. Model registry with cost attribution. Everything replayable. Everything debuggable.

THE API SURFACE

~60 routes across 15 domains.

Agents/api/v1/agents/
Ledger/api/v1/ledger/
Brain/api/v1/brain/
Evaluations/api/v1/evaluations/
Authority/api/v1/authority/
Governance/api/v1/governance/
Artifacts/api/v1/artifacts/
Runs/api/v1/runs/
Models/api/v1/models/
Events/api/v1/events/
Audit/api/v1/audit/
Tenants/api/v1/tenants/
Memory/api/v1/memory/
Sessions/api/v1/sessions/
Webhooks/api/v1/webhooks/

Versioned. Documented. Built to outlast any model vendor.

TWO ENTRY POINTS

BUILD ON CORE DIRECTLY

API-only. Full control. You architect the surface. The primitives are yours to compose. For developers and platform teams building from the infrastructure layer up.

VIEW THE API →

BUILD IN PROJECTBUILD

The governed UI surface of ProjectCore. Multi-human, multi-agent collaboration without the API-only route. Same rails. Different on-ramp.

VIEW PROJECTBUILD →
WHO BUILDS ON CORE

DEVELOPERS

Building AI-native applications that need governance from day one, not bolted on after the first incident.

SYSTEMS THINKERS

Architects who understand why the control plane matters before the surface is built.

PLATFORM TEAMS

Organizations building internal AI surfaces that need shared rails - not a different governance system per product.

TEAMS THAT CANNOT AFFORD TO GET IT WRONG

Regulated industries. High-stakes decisions. Anywhere an AI mistake has a real cost.

"The product is not the model.
The product is the system that
makes the model trustworthy."

VIEW THE API → IF YOU'RE BUILDING AT THIS LAYER - LET'S TALK →