Product · Agents
Agents that operate inside your work - not beside it
Homany was built so software agents can execute real work: they act through the same surface as the API, within the scope you set, leaving an audit trail. That's what 'agent-first, not bolt-on' means in practice.
Capabilities
What agents do here
Execute structured work
Read and update tasks, fields, and states - the same objects your team works with.
Act within scope
Permissions bound which objects and actions an agent may touch, like any teammate.
Leave a trail
Every agent action lands on the audit log - recoverable, attributable, reviewable.
Execution
How an agent moves work
Reversible by default: agents propose, humans approve what matters, then it commits.
- 1
Read
The agent reads the structured work and its current state.
- 2
Propose
It proposes the change within its permitted scope.
- 3
Approve
Anything irreversible or client-facing waits for a human approval gate.
- 4
Commit
The change commits, recorded on the audit trail.
The same surface as the API
Agents act through the operations integrations use - no weaker, separate agent path. That parity is why agents here can be trusted to execute, not just suggest.
Go deeper
Explore the agent pages
- Guide
Agent-ready operations playbook
Structure work so agents can execute safely - guardrails and a staged rollout.
Read the guide - Docs
API & MCP overview
The operational surface agents and integrations act on, and how MCP fits in.
See the API - Product
Deployment models
On-prem, local cloud, and managed tenancy - compared, with reference architectures.
Compare deployments
Put agents to work
Start free and make one workflow agent-ready, or book a session to see the agent model against yours.