Morrow by Axiom

Private intelligence,
built around you.

Morrow is a private, self-hosted personal AI agent — designed around your models, tools, memory, workflows, and privacy. Not a chatbot. Not a dashboard. An agent that is yours.

Pre-alpha · In active development

The compromise

Personal agents keep asking you to choose. Morrow is being built so you don’t have to.

  • Power or Usability
  • Cloud convenience or Privacy
  • Automation or Control
  • Flexibility or Reliability

Every one of those is a false trade. Power should not require blind trust. Autonomy without visibility is not control. Morrow is being designed to remove the “or”.

Built around you

Yours to shape — without a manual.

Morrow is designed to adapt to how you already work. Sensible defaults out of the box; depth when you want it.

Customization should feel like a setting, not a project.

Models
Bring the models you trust — choose per task.
Inference
Local by default, or an approved provider.
Tools
Add capabilities; keep them scoped.
Privacy rules
Boundaries you set, visible at every step.
Agent roles
Named agents with clear responsibilities.
Memory
Typed, scoped, and inspectable.
Interface
Desktop, web, and CLI over time.
Automation
Triggers that respect your limits.
The workspace

Watch a task move from plan to verified result.

A project, a deterministic plan, live execution, inspected files, a privacy panel, and the evidence left behind. Play it, pause it, or step through.

Quarterly review
Local modelWorkspace · ~/review
plan.jsoninspected
1{2  "goal": "Prepare the quarterly review",3  "scope": "~/review",4  "steps": 45}

Privacy

Workspace scope set to ~/review · 4 files. Nothing read yet.

Evidence

plan.json
01 / 05

Step 1 of 5: Plan the work.

Product vision preview. The execution model is in active development; this shows the intended experience, not a shipped feature.

Privacy by design

Power should not require blind trust.

Morrow is being built local-first: your data stays on your machine unless you choose otherwise — and when something leaves, you see it.

  • Local-first storage by default
  • Your choice of model and provider
  • Visible disclosure when context is shared
  • Workspace-scoped access, not whole-machine
  • No silent telemetry
  • Secret handles instead of exposed values
  • Local-only operation as a target
  • Permission boundaries you can understand

Read how privacy is designed →

Persistent agents

Not a swarm. Accountable individuals.

Named agents with a role, a model, scoped memory and tools, explicit permissions, a cost ceiling, and a history you can audit.

A multi-agent grid hands you boxes. Morrow is being designed to hand you colleagues — each one defined, bounded, and answerable for what it did.

Reliable autonomy

Act visibly. Recover honestly.

Plans persist. Work checkpoints. When something interrupts, Morrow is designed to say so — and pick up where it left off.

  1. Plan persisted

    Every step written down before it runs.

  2. Checkpoint

    Progress saved as work advances.

  3. Interrupted

    A restart, a crash, a stop — shown honestly.

  4. Recovered

    Resumes from the last good checkpoint.

  5. Verified

    Each result checked against its evidence.

Reversible actions can be rolled back. Interrupted states are shown as interrupted — never dressed up as success.

Three levels of control

Start simple. Go as deep as you like.

01

Presets

Start in minutes

Polished configurations that work without setup. Sensible models, memory, and boundaries out of the box.

02

Studio

Shape it visually

Visual control over agents, models, memory, tools, autonomy, interface, and privacy — no files required.

03

Developer mode

Go all the way down

Skills, APIs, MCP, custom providers, event hooks, and configuration files. The full surface, on purpose.

Where we are

An honest picture of what exists today.

No curated demos. This is the real state of the build, kept current.

Complete 6
  • Repository foundation
  • Product and architecture documentation
  • Local development tooling
  • Versioned contracts
  • SQLite migration foundation
  • Basic project and task persistence
In progress 4
  • Persisted task lifecycle
  • Secure workspace inspection
  • Task events and recovery
  • Connected product interface
Planned 9
  • Model providers
  • Browser and terminal tools
  • Structured memory
  • Persistent agent teams
  • Automations
  • Integrations
  • Desktop application
  • Compatibility work
  • Public benchmark suite

See the full roadmap and milestones →

Help shape the private agent
you will actually want to use.

Morrow is early, and that is the point. Join early access to influence what gets built — or follow along and watch it take shape.