Organize
Give every agent a role, a desk, and the context to do the job.
- Durable AI members with personas, memory, and skills
- Workspaces with a shared Library and Knowledge base
- Extensions add roles, workflows, and checks as data, not code
The multi-agent outcomes platform
Outcome
Many agents. One outcome.
A platform to organize, orchestrate, and operate teams of AI agents that deliver measurable results in the real world.
- Runs on Your Infrastructure
- BYO Model/Provider
- Auditable & Compliant
Durable Agents
Persistent personas with memory and skills
Multi-Agent Orchestration
Handoffs and workflows across specialized packs
Outcome-First
Measure, prove, and reproduce real results
Auditable Ledger
A verifiable record of work, decisions, and controls
Flexible & Extensible
Extensions for any domain, and BYO model or provider
Operate
Run many agents as one team, and watch them work in the Arena.
- Handoffs and workflows across specialized packs
- Live terminal sessions streamed to the Arena floor
- Bring your own model, provider, and agent CLI
Prove
Every outcome comes with the evidence that it happened.
- An append-only ledger of work, decisions, and approvals
- Checks and reviews that the author cannot sign off alone
- Deterministic replay of a workspace record
For developers
One binary. Your whole team of agents.
out is the CLI, the secure shell, and the Outcome server in one download. Install it, start a workspace, and open the Arena in your browser.
- Terminal-native agents, streamed live
- Extensions ship as plain GitHub repositories
- Files change only through recorded commands
# install
$ brew install vraison/tap/out
# start a workspace on this machine
$ out up --native
$ out status
# open the Arena
→ http://127.0.0.1:8777Start on your workstation. Scale to your organization.
Free for one host. Commercial for many hosts, central control, and SSO.