Autoheal Documentation
A self-improving software factory for the post-coding SDLC. Reliability, security, cost and supportability work that comes after the commit. Build, run, govern and continuously improve agents that run 24x7 in a governed environment, not on individual laptops.
Agent infrastructure
What every worker agent runs on.
The evaluator scores every run against downstream outcomes. The healer turns weak scores into back-tested PRs.
The graph, skills and memories every agent reads before it acts, and that Autoheal keeps improving.
Agents run 24x7 in a governed runtime, not on laptops. SaaS or BYOC, connected or airgapped.
Approval policies, an audit trail of every tool call, SSO, roles and data sovereignty.
Investigate a declared incident to an evidence-backed root cause, with approved mitigations.
Group alerts that share a cause into one prioritized event, and investigate each group.
Trace each CVE to the services that reach it and open a tested fix PR in every affected repo.
Find where your coding agents' tokens go and propose cheaper configurations, tested on your own tasks.
Define your own triggers and instructions to automate any post-coding step.
Interfaces
Work with agents from the terminal or chat.
Security & governance
Approvals, a full audit trail and control over where your data lives.
SaaS, or BYOC in your own cloud, connected or fully airgapped, with data sovereignty controls.
Require human approval before agents run sensitive tool calls.
Every tool call and approval decision, searchable and exportable.
How Autoheal secures your data across development, infrastructure, and operations.
FAQs
Where does our code and telemetry actually run?
With SaaS, Autoheal runs in an isolated tenant in Autoheal's cloud. With BYOC, the platform and its data run in your own environment, connected or airgapped. See Sovereignty for deployment and model-inference boundaries.
What happens when an agent gets something wrong?
Findings include their supporting evidence so your team can review them. Governance policies can require approval before sensitive tool calls. The self-improvement loop evaluates outcomes and proposes changes for an engineer to approve.
Can we bring our own model?
Agents use models from your organization's approved list. Under BYOC, inference can use your own provider account or model endpoint. See Build Your Own and Sovereignty.
How is this different from a coding assistant?
Autoheal handles the reliability, security, cost and supportability work after a commit. Agents run in a shared, governed environment and read shared context. Autoheal works alongside the coding assistants your developers already use. See How It Works.
How is it priced?
Contact the Autoheal team for current pricing and deployment options for your team.