Keep the intelligence you already use.
Claude, ChatGPT, Gemini, your own API key, or ODEI credits. The model layer remains replaceable while the personal layer becomes yours.
Announcement coming soon
Create your personal AI agent.Your own World Model, governance, and execution loop in under 5 minutes.
Bring Claude, ChatGPT, or Gemini. ODEI turns the intelligence you already use into a governed personal runtime that remembers, acts, verifies, and improves across sessions instead of starting from zero.
This is the runtime preview and founding cohort access lane. The public launch is not open yet, but the exact onboarding contract is already live.
Your own World Model
Observe → Decide → Act → Verify → Evolve
Governed execution
Receipts on every action
Portable model layer
How onboarding works
Full product thesis
Turn inboxes and tools into one graph.
Email, calendar, documents, tasks, CRM, and files stop living as disconnected integrations and start forming your World Model.
Define what the agent can do.
Approvals, action scopes, quiet hours, budget limits, and sensitive boundaries are defined before the loop is allowed to act.
Start from real work, not a demo.
Inbox triage, follow-ups, operating briefs, pipeline monitoring, or a custom mission that matters to your actual operating cadence.
Public announcement is coming soon. This surface shows the real onboarding contract now, and the intake below is the controlled lane for the first founding cohort.
Preview runtime presets
Sharable before activation opens
Inbox triage, weekly operating briefs, and governed follow-ups built around your real operating cadence.
Silent thread detection, next-touch preparation, and deal momentum across email, calendar, and CRM context.
Watch a fixed perimeter, update the World Model, and surface receipt-backed briefs when something materially changes.
Under 5 minutes means the first runtime boundary is defined quickly: model lane, world sources, governance mode, and first mission. Deeper data expansion happens after activation.
Best first fit
Start with repeated high-value work
Best for founders who need inbox triage, investor follow-ups, and weekly operating briefs to keep moving without losing context.
First mission: Prepare my weekly operating brief and surface the follow-ups that actually matter.
Best for operators running pipelines, counterparties, and silent threads where timing, context, and next-touch quality drive revenue.
First mission: Track silent threads and draft the next follow-up with deal context already loaded.
Best for people who monitor a fixed perimeter and need governed briefs when something materially changes.
First mission: Watch my research perimeter and issue a receipt-backed brief when the landscape moves.
Best for small teams that need one shared World Model, governed operator lanes, and receipts across execution.
First mission: Stand up one shared runtime with approvals, team context, and a first operating cadence.
The first cohort is not for everyone at once. It is for operators with one repeatable mission where memory, governance, and execution change the outcome immediately.
Founding cohort operating contract
Invite-first and execution-focused
Founders, deal operators, research operators, and small teams with one mission where governed agency changes the outcome immediately.
One model lane, three world sources or fewer, one mission with a real cadence, and a governance mode that defines what the runtime may do.
Deterministic intake envelope, provisioning capsule, runtime scope, governance boundary, and the first mission queue instead of a vague waitlist state.
Intake returns immediately, runtime scoping and governance review happen in the first operator review window, and activation starts only after reviewed provisioning.
What “under 5 minutes” means
Exact onboarding boundary
Choose model, governance, world sources, execution surfaces, and the first mission, then carry that exact runtime shape in a preview URL.
Before submission, the app exposes the same structured payload that the intake API will receive. Nothing is hidden behind the UI.
After submit, the API immediately returns intakeId, lane, receivedAt, and consentAt.
The app then renders the provisioning capsule from that same runtime contract, so the first handoff artifact exists immediately.
This is the promise: you leave with a bounded runtime contract and the first provisioning artifact, not a vague waitlist form.
What to bring before you submit
Start narrow on purpose
Pick the intelligence layer first: your subscription workflow, your own API key, or ODEI credits. The model should stay replaceable.
Start with three sources that already matter: email, calendar, docs, CRM, or tasks. The goal is a useful first graph, not a perfect one.
Define one mission you already repeat every week: inbox triage, follow-ups, operating brief, or research monitoring. This keeps the first runtime measurable.
Founding-cohort onboarding is opinionated by design: one clear runtime, one governance boundary, and one first mission that can be verified with receipts.
Pricing structure
Human billing · machine usage
Setup
Initial provisioning
We map the first World Model, connect tools, define approvals, and configure the missions that should run from day one.
Monthly runtime
Persistent agency
Memory, governance, receipt storage, hosted loop execution, and continuity across sessions live here.
Usage-based execution
Heavy actions scale with real work
Browser actions, large model calls, paid routes, and external API spend scale with execution rather than idle seats.
Commercial shape is fixed now: setup, runtime, and metered execution. Public self-serve billing will open with the launch announcement.
Account and activation boundary
Identity, entitlements, and billing readiness
The preview is public, but runtime activation still starts from an intake ID, verified operator details, and a reviewed mission boundary.
Setup, runtime, and usage are already defined. Founding cohort onboarding captures billing readiness now, even while public checkout remains closed.
Activation is not a vague “approved” state. It yields a World Model workspace, governance profile, first mission queue, and receipt surface.
What happens after submit
Activation contract
You immediately get the intake envelope.
The API returns the deterministic intake envelope first: intake ID, lane, source, receivedAt, consentAt, a normalized summary, and the first provisioning trail. The app then renders the provisioning capsule from the same submitted runtime contract.
We shape the first agent around your actual world.
World sources, model layer, execution surfaces, and first mission are translated into the initial runtime scope instead of a generic assistant template.
Approvals and boundaries are locked before activation.
Quiet hours, action permissions, and sensitive lanes are confirmed so the first loop launches with explicit control and clear receipts.
You receive the first runtime, not just another login.
The output is a working personal runtime: World Model, governance profile, receipt surface, and the first mission queued to run.
What the first runtime includes
Provisioning schema
The first output is not a vague waitlist entry. You get an intake ID and a machine-readable capsule bound to your exact onboarding contract.
The runtime is scoped around your chosen sources, your execution surfaces, and the approval rules that define what the loop may do.
The first runtime ships with one live mission queued and a receipt trail you can inspect as the agent observes, decides, acts, and verifies.
The business promise is explicit: your first runtime should already feel like a governed operator, not like another empty dashboard.