Macro 3D Playbook court with an AI agent paused at a human approval gate.
How It Works

Start with what you have.

Bring a task to automate or a product you have already built. We can review the existing work, plan a new workflow, or build and test an agent. We agree on the scope before you commit.

  • Owner Named
  • Protected action Held
  • Evidence Attached

When to call us

Built something. Need a second pair of eyes?

You can keep building while we help you check the parts you cannot confidently verify.

  • You built a product with AI and want an engineer to review it before a customer pilot.
  • A feature keeps breaking despite repeated fixes.
  • A customer asks for an integration or security review, and you need to know what you can commit to.
See how a technical review works →

Starting from an idea? Build one useful AI task. Already running a system? Explore ongoing support.

An ivory model of a connected system, with one part lifted into an inspection frame while its neighbors remain in place.
Concept illustration Inspect the agreed part. Keep the rest in view.

Choose where to start

Plan the work, build it, then support it.

Map helps you plan the work. Build creates a system your team owns. After launch, Control provides Managed AI Operations from $900/month for a standard-risk environment. Control includes Map. New workflows and integrations are quoted separately through Build. The supported public source distribution is $0 / MIT.

  1. $0 browser-local starter · Account workspace · pricing at launch

    Map

    Define the workflow

    A workflow plan you can update, share, and use to guide the build.

    Explore Map
  2. Scoped implementation

    Build

    Build and test the agent

    A tested system with clear permissions and instructions your team keeps.

    Explore Agent Foundation
  3. Managed AI Operations · From $900/month · includes Map

    Control

    Keep the live system working

    A managed system with a workflow plan, work history, recovery instructions, and regular reviews.

    Explore Control
Verified field result

Automation prepared the evidence. Human judgment still decided.

The system gathered evidence for reviewers. A person still decided whether to approve or reject each submission.

49/ 50
49 of 50 selected cases produced usable evidence packets for human decision.
Workflow
Marketplace template review
Receipt
#FR-2026-01
Verified
May–June 2026
Changes to external systems
0

Map before booking

Try planning the task before you commit.

Choose a task. Map its steps, the people responsible, and where AI could help. The public canvas lets you draft a plan without connecting to live systems.

01 · Trigger Request arrives

Person, system, or agent

02 · Boundary Prepare, then wait

AI drafts · a person approves

03 · Proof Record the result

What happened · who acts next