Orchestration

Governance-aware AI agent coordination for software teams. Plan, build, verify, and deliver with structural integrity.

Get Started

What is Orchestration?

Orchestration is a framework for AI-assisted software development that ensures every code change is planned in context, validated against architecture, and delivered with full traceability. It coordinates AI agents through a structured lifecycle — from request intake through planning, execution, validation, and delivery — with governance gates at every step.

Instead of ad-hoc AI coding, Orchestration gives your agents persistent memory across sessions: what your system looks like, what work is in progress, and what decisions have been made. Your agents build on prior work instead of starting from scratch.

The Triad: Three Services, Complete Memory

Structural Memory

AKL

Architecture Knowledge Layer. Stores your product breakdown structure, component registry, architectural decision records (ADRs), conventions, and feature traceability. Agents know what your system looks like before making changes.

Operational Memory

Doit

Work tracking and issue management. Manages epics, tasks, dependencies, and lessons learned. Agents know what work is in progress, what's blocked, and what mistakes to avoid repeating.

Communication Memory

Herald

Inter-agent messaging and coordination. Enables agents to send typed messages (DO, ASK, TELL), signal progress, and broadcast architectural changes across project boundaries.

What You Get

Governance Gates

7 automated gates from orientation through Definition of Done. Structural boundaries, feature preservation, and behavioral tests enforced at every step.

Agent Topology

10 specialized agent roles — from Architecture to Verification — each with defined capabilities, constraints, and tool access.

9-Phase Lifecycle

Every request follows a structured path: intake, orientation, planning, execution, validation, reconciliation, verification, and delivery.

Continuous Improvement

Lessons learned from failures are recorded and consulted before every task. Your agents get smarter over time, not just faster.

Risk Classification

4-level risk assessment (LOW to CRITICAL) determines approval requirements and governance depth automatically.

Product Traceability

User needs trace to features, features to acceptance criteria, criteria to test specs. Nothing falls through the cracks.

How It Works

1

Register

Use your invite code to create an account. You'll receive API keys for all three Triad services.

2

Install scaffolding

Claude reads the setup instructions and installs configuration files, governance prompts, and slash commands into your project.

3

Connect Triad

MCP server configuration connects Claude to AKL, Doit, and Herald with your API keys.

4

Start building

Run /orc-orchestrate for full lifecycle governance, or /orc-create-epic to plan work packages.

Ready to orchestrate?

You'll need an invite code from your administrator.

Get Started