MCP-Native Platform

AI agents that ship your roadmap

The orchestration layer that grounds AI agents in your product backlog, team standards, and codebase knowledge graph — built for engineering teams who need rigour, not magic boxes.

One Platform, Three Pillars

Grounded Backlog, Skills Governance, and Knowledge Graph — the complete orchestration stack for engineering teams using AI agents.

Grounded Backlog

AI that ships your roadmap

Your epics, features, and tasks with acceptance criteria — AI agents work against your actual product backlog, not hallucinated requirements. Every session ties back to a business outcome, and nothing ships without passing human-defined gates.

8/8 planning tasks passed
ctrl.sessions ctrl.criteria ctrl.evidence ctrl.gates
Skills Governance

Your team's AI playbooks, centrally managed

Internal enterprise skills that encode your team's conventions, architecture decisions, and coding standards. Every AI agent follows the same playbooks — not ad-hoc prompts scattered across individual setups.

13.4% of ClawHub skills flagged for security issues
ctrl.skills ctrl.criteria
Knowledge Graph

Your codebase, understood

Knowledge graph and ontology that gives AI agents structural understanding of your code — relationships between files, modules, and APIs. Fewer tokens wasted on context gathering, better results from every interaction.

Significant token reduction
ctrl.knowledge ctrl.stacks

Get Started in 3 Steps

One CLI command. Zero config files. Full orchestration in minutes.

1

Add MCP Server

Run one command to connect Claude Code to aictrl.dev via the Model Context Protocol.

2

Sign In

Claude Code opens your browser for Google sign-in. No API keys, no tokens to manage.

3

Start Building

Your knowledge graph provides context, skills encode your standards, and every session tracks against your backlog.

Terminal
$ claude mcp add --transport http aictrl "https://app.aictrl.dev/mcp"

Everything You Need to Ship with Confidence

Six capabilities across three pillars that turn AI chaos into engineering rigour.

ctrl.criteria

Acceptance Criteria

Define what "done" means. Features aren't complete until criteria are verified with evidence.

ctrl.evidence

Evidence Trail

Screenshots, test outputs, commit hashes — all captured automatically as proof of agent work.

ctrl.gates

Approval Gates

Require human sign-off for sensitive operations. Never deploy without explicit review.

ctrl.knowledge

Knowledge Graph

Ontology-driven structural understanding of your codebase — relationships between files, modules, and APIs.

ctrl.skills

Skills Governance

Internal enterprise playbooks that encode your team's conventions. Centrally managed, not ad-hoc prompts.

ctrl.analytics

Analytics

Session metrics, team performance, and trend analysis. Data-driven engineering decisions.

Not Another Code Generator

We build workflow orchestration, not magic boxes.

Lovable / Bolt Cursor GitHub Copilot aictrl.dev
Target Audience Non-technical Developers Developers Engineering Teams
Verification None None None Acceptance Criteria
Audit Trail No No Logs only Full Evidence
Long Sessions No Single Context Single Context Checkpoints
Human Gates No Manual No Built-in
Team Dashboard No No Basic Real-time
Knowledge Graph No No No Ontology-backed
Skills Governance No Rules files No Enterprise-managed

One Platform, Three Perspectives

Whether you lead the org, manage the team, or write the code — aictrl has your back.

Governance, ROI, and Business Alignment

Your board asks about AI ROI. Your compliance team worries about ungoverned agents. aictrl gives you the evidence to answer both — every AI session tied to a business outcome, with full governance and cost visibility.

  • Every AI session tied to a business outcome
  • Organisational AI consistency at scale
  • Reduced AI context costs via knowledge graph
  • Approval gates prevent unsanctioned deployments

Designed for

Prove AI ROI to the board with real productivity data and the governance controls leadership demands.

Standards, Quality, and Team Productivity

You manage 5-50 engineers using AI agents daily. You need to know what's shipping, what's stuck, and whether standards are being met — without reading every PR.

  • Acceptance criteria enforced before anything ships
  • New engineers productive on day one
  • Any engineer can understand any repo
  • Evidence trail makes debugging faster when things go wrong

Designed for

Stop worrying about what AI agents are doing unsupervised. Check the dashboard — if criteria are green, move on.

Context, Patterns, and Speed

You're already using Claude Code. aictrl makes your AI sessions smarter — the knowledge graph finds the right files, skills give you proven patterns, and your progress persists across context resets.

  • Context persists across session resets
  • Proven patterns for every common task
  • Agent finds the right files immediately
  • Zero friction — one CLI command and you're connected

Designed for

Add aictrl in 30 seconds and get full codebase context plus your team's best patterns in every Claude session.

Frequently Asked Questions

Quick answers to common questions about aictrl.dev.

What is aictrl.dev?
aictrl.dev is the orchestration layer that grounds AI agents in your product backlog, team standards, and codebase. It is not a code generator — it coordinates AI coding agents through three pillars: a Grounded Backlog with acceptance criteria and human approval gates, Skills Governance for centrally managed team playbooks, and a Knowledge Graph for structural codebase understanding. Think of it as the missing coordination layer between your AI agents and your engineering process.
How do the three pillars work together?
The Knowledge Graph provides the foundation — structural understanding of your codebase that makes everything above it smarter. Skills Governance encodes how your team works into reusable, versioned playbooks. The Grounded Backlog ties AI work to business outcomes with acceptance criteria and human approval gates. Each layer amplifies the one above it: the knowledge graph informs skills with codebase context, and skills power the backlog with team standards. Cross-cutting analytics span all three layers.
How does aictrl work with Claude Code?
aictrl connects to Claude Code via the Model Context Protocol (MCP). Run one command (claude mcp add --transport http aictrl "https://app.aictrl.dev/mcp"), sign in with Google, and Claude automatically receives codebase context from the knowledge graph, follows your team's skills, and tracks sessions against your backlog. No manual configuration needed — just one CLI command to connect.
Is there a free trial?
We're currently offering early access. Contact us or sign up to get started with full access to all features.
What about security and data privacy?
aictrl captures session metadata, acceptance criteria status, and evidence files — your source code is never stored on our servers. Skills Governance ensures your team's AI playbooks are internal and enterprise-managed, not pulled from unvetted public sources where a ClawHub audit found 13.4% had security issues. All data is encrypted at rest and in transit. See our Privacy Policy for full details.

Ground Your AI Agents in What Matters

Three pillars. One platform. Start shipping your roadmap.

Get Early Access

Sign up for early access. No credit card required.