Introducing IAS: Why We Built an Operating System for AI-Assisted Development

by Niklas Mencke

AI agents are reshaping how software gets built. Engineers use them to write code, review pull requests, debug issues, and even plan architectures. But there's a gap between running an AI agent and running a team of AI agents alongside humans — reliably, transparently, and at scale.

That gap is what IAS fills.

The problem

Most teams using AI for development today face the same pattern: an engineer prompts an agent, the agent generates code, the engineer reviews it and merges. It works for individual tasks. But it breaks down when you need:

• Multiple agents working in parallel across different parts of a codebase
• Non-technical stakeholders to have visibility into what's happening
• Audit trails that show not just what changed, but why
• Guardrails that prevent agents from going off track without blocking the entire pipeline

The tooling doesn't exist yet. Teams cobble together scripts, custom wrappers, and Slack notifications. It's fragile and opaque.

What IAS does

IAS is three things:

1. An agent framework — a structured way to deploy AI agents that operate on your codebase. Git-native, sandboxed, auditable. Agents work in branches, submit pull requests, and follow your existing review workflow.

2. A command center — a control plane where you can monitor agent activity, steer work, make decisions, and review outcomes. Think of it as the dashboard for your AI-assisted development pipeline.

3. A context lake — a unified knowledge layer that aggregates project context (docs, tickets, conversations, code) so agents have the information they need to act intelligently.

Together, these form an operating system for AI-assisted development.

Who it's for

We built IAS for three audiences:

AI engineers who want to run agents on their codebases without losing control. They care about auditability, Git-native workflows, and the ability to intervene when something goes wrong.

Product managers who want to steer AI-assisted development without needing to read code. They care about visibility, decision routing, and knowing that nothing ships without human approval.

Decision makers — CEOs and engineering leaders who need visibility, governance, and control over AI-assisted development across their organization.

What's next

We're actively building and shipping. If you're interested in running IAS on your codebase, get started or reach out — we'd love to hear what you're building.