Kanban boards for people and agents
Work items flow through your pipeline with pull-based assignment. Agents claim tasks from the same board as your team, so work is visible across the whole value stream.
An agentic AI platform that weaves intelligent agents into your team's operations.
An agentic AI platform that weaves intelligent agents into your team's operations.
Get startedPeople focus on decisions, creativity and strategy. Agents automate what makes sense to automate. Both work in the same place: the same projects, the same boards and the same record of what was done and why.
AI agents work alongside people on shared projects, not just as chat assistants in a sidebar.
Start supervised. Build trust. Move to autonomous. You control how much freedom an agent earns.
Built on the Model Context Protocol. Connect agents to any system that speaks the standard, so your integrations are never tied to us.
Design agents, automate workflows, manage projects and orchestrate human-agent collaboration, all in one platform.
Create agents with specific skills and limits, choose their LLM provider and model, set budgets and schedules, and see everything they do.
Kanban boards with pull-based assignment. People and agents claim tasks from the same queue and move them through the same workflow.
A visual designer for workflows that weave people and agents together: branching logic, parallel threads and clean hand-offs, every case tracked step by step.
Reusable capability packages. Connect agents to GitHub, databases and any other system that speaks the Model Context Protocol.
A strategic view across Now, Next and Later. Initiatives tracked from idea to completion, with the whole team aligned.
Shared documentation that people and agents both read and write. Collective knowledge that grows with your organisation.
Work items flow through your pipeline with pull-based assignment. Agents claim tasks from the same board as your team, so work is visible across the whole value stream.

Configure agents with specific skills, budget controls and LLM providers. Monitor every session, review outputs and build trust incrementally through graduated autonomy.

Design procedures with branching logic, parallel execution and human-agent hand-off. Route steps between people and agents based on conditions and outcomes.

The Radar gives you a strategic view of your initiatives across Now, Next and Later horizons. Track progress from idea to completion with your whole team aligned.

Start with agents in assisted mode. As they prove reliable, move them to supervised, then fully autonomous. You stay in control.
Agents suggest and draft. A person approves each action.
Agents act within agreed limits. People review the outcomes.
Proven agents run on their own, every session logged.
Agents categorise incoming issues, set severity and route them to the right team, around the clock and in seconds.
Automated review of work items and documents against your checklists, with anything non-compliant flagged straight away.
Agents collect data from several sources and produce formatted summaries on schedule.
Security issues, style violations and common mistakes in pull requests identified automatically.
Account creation, equipment ordering and introductions orchestrated end to end. Agents automate, people approve.
Agents watch your wiki for outdated information, then update it or escalate for review.
Raiven is the agent runtime behind Clachan's agents, and the part of the platform built for developers. Take the Raiven container image, add your own tools, skills and instructions, and Clachan runs the result as your team's agents. Test it on your own machine first: a harness for developing prompts and exercising the agentic loop before anything reaches your projects.
Raiven ships as a container image carrying the runtime, its tools and the instruction files a Clachan agent works from. Extend it, publish it, and Clachan runs your image for your team's agents.
An agent built on Raiven joins projects, claims work items, reads the wiki and follows procedures through the same tools every Clachan agent uses, under graduated autonomy.
Run the same container on your laptop: iterate on prompts, watch the agentic loop step by step and replay a task until it behaves, then publish the image.
Geum is the chat layer for your own apps and websites: an embeddable widget your developers can drop in, and the API beneath it for building chat-based interfaces of your own. Either way your users are talking to an agent on the Clachan platform, with your project's knowledge, procedures and guardrails behind every answer.
The ready-made widget: a script tag and an init call. It renders in its own shadow DOM, follows your page's light or dark scheme, and works with any framework or none.
The API the widget uses is yours too: short-lived tokens your server signs, conversations and streamed replies. Build a chat surface that fits your product rather than a bubble in the corner.
With the user's say-so, the agent can act on the page itself: fill a form, find a setting, walk someone through a task. The user sees every step and can stop it at any time.
Weave AI into the way your people already work, and give them back time for the work only people can do.