README.md
August 17, 2026 · View on GitHub
Shared infrastructure for accelerating AI-native research
Documentation · Launch DAM · Join the Waitlist
Why DAM?
- ☁️ Runs in the cloud. Agents execute continuously in the cloud and keep running after you close your laptop or go offline.
- 🔐 Isolated execution. Each agent runs in an isolated container with all access routed through a policy-enforced gateway.
- 🔑 Zero trust credentials. Connect agents to your tools without exposing credentials to the runtime.
- 👥 Built for teams. Collaborate in Slack and run agents on schedules for recurring workflows.
Ways to Use DAM
| Mode | Description |
|---|---|
| Web UI | Chat with your agent, stream its terminal, and manage files — all from the browser. |
| CLI | Create agents, attach to live sessions, and manage instances from your local terminal. |
| Slack | Message your agent from Slack threads. Teammates interact with their own credentials. |
| Schedules | Run agents on a recurring timer — daily code reviews, nightly audits, continuous monitoring. |
Supported Agent Harnesses
| Harness | Description |
|---|---|
| Claude Code | Reasoning-first assistant for complex coding tasks. |
| Pi Agent | Multi-provider coding harness across leading LLMs. |
| Bob | Enterprise coding assistant for IBM workflows. |
| Codex | Execution-first system for end-to-end coding tasks. |
Bring your own harness — any runtime compatible with ACP can run on DAM.
Get Started
Head to ibm.biz/dam-agents, create an instance from a template, and start chatting.
See the documentation for quickstarts, core concepts, integration guides, and use cases.
Developing DAM locally
For contributors working on the DAM platform itself.
Prerequisites
- mise
- Docker-compatible runtime (Docker Desktop, Rancher Desktop, Colima, etc.) -- note that Podman is not supported
- macOS or Linux
- on Linux, either:
- install QEMU if you want to run k3s in a VM (default), or
- set environment variable
IS_SANDBOX=1if you want to operate directly in the current OS, typically if it is already a VM
Local Setup
git clone https://github.com/dam-agents/dam && cd dam
mise install
mise run cluster:install
Open localhost:4444 and log in with:
username: dev
password: dev
Create an instance from a template and start chatting with your agent.
See work process for the contributor workflow.