nomosV0.1 · ALPHA
R00Roadmap

One operator today. An AI workforce under authority.

Five stages on one object model. Each stage is a usable product on its own, and none of them requires re-architecting the layer beneath it.

01 · Now
Personal MVP
02 · Next
Knowledge layer
03 · Defined
Business MVP
04 · Defined
Enterprise
05 · Intent
Autonomous work
001Stage specification

Each stage adds scope, not a second product.

The primitives introduced early — sessions, memory records, grants — are the same ones an enterprise deployment runs on. Status is stated plainly: in development, defined, or intent.

STAGE 01IN DEVELOPMENT
Personal MVPOne operator runs all their AI work through a single workspace: organized projects, every model reachable, history that persists.projects · sessionsrouting · unified history
STAGE 02NEXT
Knowledge layerWork stops restarting. Decisions, sources, and constraints carry across projects and survive a change of model.memory records · lineagesemantic retrieval
STAGE 03DEFINED
Business MVPA team shares context without sharing everything. Roles, scopes, and shared projects replace forwarded screenshots.teams · roles · departmentsshared knowledge scope
STAGE 04DEFINED
EnterpriseAI becomes deployable in a regulated organization: standing policy, named approvers, audit export, systems of record in scope.policy engine · audit exporterp/crm connectors · self-host
STAGE 05INTENT
Autonomous workAgents hold standing authority for defined scopes and report against it — an AI workforce that can be audited like a human one.standing grants · agent registrydelegation · assurance
002Stage 01 · In development

What we are building right now.

A complete personal operating system — narrow in audience, complete in structure. If a single operator can't run their entire AI workflow through Nomos, nothing later in the sequence matters.

alpha access is limited and hands-on · we work directly with early operators
BUILD SCOPE · STAGE 01V0.1 ALPHA
Projects & sessions
structure for work, resumable
Multi-model routing
cloud + local, per task class
Unified history
searchable across models
Knowledge attachment
sources scoped to projects
Persistent context
carried between sessions
Tool & MCP bridge
actions under explicit grant
Authority primitives
grants + audit from day one
Agent execution
human in the loop by default
Governance is in stage 01 deliberately. Retrofitting authority into a system that shipped without it is the mistake we are designing around.
003Dependency order

Why the order can't be rearranged.

Most AI products sell autonomy before they can prove authority. The sequence here is the reverse: structure, then memory, then shared scope, then policy, and only then standing autonomy.

Structured workgives memory something to attach to
Durable memorymakes shared context worth sharing
Teams & scopesgive policy something to govern
Policy & auditmake autonomy safe to grant
Standing authorityan AI workforce that can be held accountable
004Enterprise · Forward scope

Where a personal operating system becomes an organizational one.

The same primitives extend outward: business knowledge instead of personal notes, departments instead of projects, policy instead of preference.

status: architecture defined · stages 03–05 · not yet released
01
Business knowledge
SOPs, contracts, reports
02
Systems of record
ERP, CRM, ticketing
03
Policy engine
constraints, approvals
04
Departments
roles, teams, scopes
05
Audit & assurance
who, what, when, why
06
AI workforce
agents under authority
005How we sequence
Depth before breadth.
A complete operating system for one person beats a shallow one for an enterprise. Stage 01 ships whole.
No stage invalidates the last.
Every stage runs on the same object model, so early adopters are never migrated onto a different product.
Stated, not implied.
Anything not yet built is labelled defined or intent. We would rather be held to the sequence than oversell the present.

Come in at stage 01.

Alpha operators shape what stage 02 becomes. Access is limited, hands-on, and free during alpha.

For investors

The sequence is the thesis: own the layer that holds work, memory, and authority while the models underneath commoditize.

Read the thesis