Live now ProjectLedger

Memory and governance for AI colleagues.

Project0 is the substrate behind multi-human, multi-agent workflows. Persistent memory across sessions, teams, and time. Sealed decisions. Durable memory. Declared authority.

An AI co-worker is allowed to be wrong. It is not allowed to be ungoverned.

Product preview
Promote
human_led scanner.gaming.v2
2s ago
0x7a3f9c81b2c14e09a85d27f4ed8be...
Promote
human_in_loop desk.equities.pmA
18s ago
0xa2f81bd09e734c50ff4123de9176...
Promote
agent_autonomous ops.studio.creator
1m ago
0xc8e02a7fb31994ec70e5912cb831...
Append-only · Tamper-evident · KMS-signed ProjectLedger ↗

Built for Vertical surfaces built across gaming· fintech· education· healthcare· AI infrastructure

Our Substrate: Ready to Build?

License the engine that built all six.

Engine
Core
Enterprise API available

The API-first control plane behind everything Project0 builds. Run agents with declared authority, open bounded evaluation windows, promote outcomes to an append-only ledger, seal everything into verifiable artifacts.

Talk to integration →
Workspace
Build
API available

The governed workspace for multi-human, multi-agent workflows. Persistent memory across sessions. Authority declared at every decision. Built for teams where humans and AI colleagues share the same operational surface.

Get a demo →
Compliance
ProjectLedger
Live in production

Pass your audits with proof, not promises. Sealed decision records, KMS-signed hashes, independently verifiable. Built for regulated workflows where evidence matters.

View ProjectLedger →
Services

We embed and ship multi-human, multi-agent systems on the substrate.

Founder-led with senior bench depth across AAA gaming, infrastructure, and platform engineering. The category almost nobody else is shipping in. Selective engagements.

01

What we build

Custom agents on Project0 Core. Domain-specific platforms for your industry. End-to-end systems with sealed governance from day one.

02

How we work

30 to 60 day pilot on a single use case. 18-month Founding Partner engagement default. Weekly cadence, founder-led delivery.

03

Who you get

Founder plus senior bench across AAA gaming, infrastructure, and platform engineering. Pricing reflects scope and value, not seat count.

Verification model

What a sealed-decision check returns.

Every decision sealed by Project0 lands in an append-only chain and is KMS-signed by the chain anchor. Below is the shape of a verification response — a sample record, not a live lookup.

Sample response Fields returned by a chain verification.
Commitment hash0x7a3f9c81b2c14e09a85d27f4ed8becd9176af2c401de8b934e7af0a217cd9af3
Prior hash0x2f81bd09e734c50ff4123de9176af2c401de8b934e7af0a217cd9af37a3f9c81
Authority modehuman_led
KMS key versionchain-anchor / v3
Server timestampRFC 3339, set on write
This panel shows the shape of a verification response. It does not check a hash against a ledger. Production verification runs server-side through Core’s chain-verify endpoint, authenticated and scoped to your own project’s chain.

One hash is the start. The full ledger is the value.

01 · One decision
The hash above.

A single sealed commitment. Hashed on write and KMS-signed by the chain anchor. Verifiable by any party you issue a tenant credential to, against the append-only chain.

Authenticated, tenant-scoped
02 · One project's history
Every Promote(), every gate, every authority change.

A replayable timeline of every commitment a project has ever made. Searchable by actor, by authority mode, by evidence manifest. Days, weeks, months of operational record, reconstructible on demand.

Append-only · never overwritten
03 · Institutional memory
Decisions outlive the conversation that produced them.

Years of sealed evidence across teams, sessions, agents, and humans. Audit trails built to support the evidence SR 11-7, NIST AI RMF and EU AI Act work demands, without needing to reconstruct what happened months ago. Mapping that evidence to your control framework stays with your counsel.

Open user retention policies
Why this, why now
01The line

An AI co-worker is allowed to be wrong. It is not allowed to be ungoverned.

This is the line Project0 draws in software, for the specific case almost nobody else is solving today: humans and agents working as a single team across the same memory, the same authority surface, the same ledger.

Memory persists across sessions and across actors (humans, agents, tools), so the next move inherits everything the last move knew. Authority is declared at every commitment, so a buyer or regulator or auditor knows which actor was in which mode. The Promote() boundary makes every sealed decision a tamper-evident artifact. The Ledger ensures evidence outlives the conversation that produced it.

Wrong is allowed. Ungoverned is not.

02The reality, and the gap

AI colleagues are already in production. The infrastructure to govern them is not.

LLMs and agents are already making decisions in your production pipeline. Cursor ships code that lands in main. ChatGPT drafts live-ops checklists. LLM-as-judge moderates content at scale. Whether or not your team officially adopted them, AI colleagues are running against your production data today.

When one of those actions lands (a deploy gate fires, a trade clears), there is no sealed record of who decided what. No declared authority. No way to verify the claim against the artifact.

The regulatory floor is rising fast: SR 11-7 governs model risk for banks. NIST AI RMF defines the bar for federal contractors. The EU AI Act enforces high-risk obligations from August 2026. Companies running agents on undocumented authority are building latent liability with every decision they let pass unrecorded.

This is not a model gap. It is a substrate gap.

How to deploy Project0

License the engine. Buy a platform. Hire the team.

Three paths. One substrate. Rails are boring until the train is moving at 300 km/h.

Project0 · Built on Project0 infrastructure

Prefer the long version? Read the story ↗