Skip to main content
    ● In production today · Part of the Forge portfolio · Powered by Reign
    Dark Factory — Governed Autonomous SDLC

    The agents were never the hard part.
    Governing them at scale is.

    Governed autonomous development.
    Human control. Enterprise evidence.
    Dark Factory is a licensed platform for governed autonomous software development. It integrates with the tools your teams already use, runs autonomous AI agents with a human approval gate before every merge, and delivers tamper-evident, audit-ready evidence powered by Reign.
    Request a 30-minute briefingTake the 3-min readiness assessment →
    5–7×
    Effective engineering output
    ~21 hours
    Median cycle time
    0 reverts
    across 922 merges
    100%
    Human-approved merges
    Dark Factory · Supervisor ConsoleLIVEWORKFloorQueue 27CREWCrew 6OBSERVEActivityCost $5.80/hrOPERATEReleasesAuditprod-floor · ip-10-0-2-14 · supervisor up 3d 4h · tick 30sAll 18Running 3Blocked 4Done 1READY1reign-d1waiting for slotgt-d1RUNNING3reign-x4auto-review · cautiousgt-x4reign-cqnbuild + unit testsgt-c1reign-a7security scangt-a7BLOCKED4reign-b2dep upgrade conflictgt-b2reign-k9needs human reviewgt-k9+2 moreDONE1reign-p0merged to main ✓gt-p0BEAD SPECreign-x4typetaskpriorityP2weight5laneauto-reviewworkflowdefaultexternal_refgitlab !105Open MR ↗● awaiting human approval✓ Approve & mergehuman-in-the-loop · sealed to ledgerhealthy · queue 12 · 4 workers · $5.80/hr · auto-task at 2/2 · reign-d1 waiting for slot
    ● LIVE — Real-time view of the Dark Factory Supervisor Console. Every agent action is observable. Every merge requires explicit human approval. All activity is recorded for audit.
    Why now

    Everyone has agents.
    Almost no one can govern them.

    The real challenge isn't building with agents — it's running them at production scale with control, evidence, and compliance.
    Pilot purgatory
    Most teams can't move agentic work into production while meeting governance and audit requirements.
    Tool sprawl & cost
    Enterprises run 300+ SaaS tools at under 50% utilization — creating massive hidden spend.
    Regulation is here
    OSFI E-23 and the EU AI Act now require runtime governance and evidence for autonomous systems.
    How it works

    File a ticket. Agents build it.
    You approve. It ships.

    Work is pulled from the trackers you already use. A Supervisor routes each ticket to an AI worker running in an isolated sandbox. The worker completes the change and runs every quality gate. A finished merge request waits for one required human approval before shipping — with full evidence captured by Reign.
    THE BACKLOGTHE FLOOR MANAGERTHE CREWTHE RESULTWhere work comes fromIssue trackertickets & bugsGit providerPRs & reviewsProject boardepics & tasksAny sourcevia open APIsany tracker, one interfaceSUPERVISORone always-on controllerBacklog Interfaceone common plug forevery tracker —finds ready workWorker Managerclaims a ticket,starts a worker,retries on failureWorker 1AI in a DockersandboxWorker 2AI in a DockersandboxWorker 3AI in a DockersandboxMerge Requestcode written + tested,all quality gates passed,AI pre-reviewed.Ready for a human.Human approvesthe safety gateMerged to mainticket auto-closes
    From tracker to production — each ticket goes to an AI worker in its own isolated Docker sandbox (reproducible, contained, zero blast radius); the finished change arrives as a merge request and waits on a human before it merges. Every step governed by Reign, with tamper-evident evidence.
    Velocity
    Ship in hours, not weeks
    Control
    Human approval on every change
    Cost
    Integrate now. Consolidate later.
    The stakes

    Two paths forward.

    Ungoverned
    Velocity you can't prove. Growing tool sprawl. Rising compliance risk. Audit gaps that only get wider.
    →
    Governed
    5–7× output with full human oversight. Tamper-evident evidence on every change. Lower tooling spend over time.
    Open by design

    Your stack, governed.

    Your choice of coding agents — flexible, independent and open, never locked into one vendor, model or cloud. Dark Factory integrates with the toolchain you already run via open APIs, webhooks and MCP — and governs it end to end with Reign.
    Coding agents
    CursorClaude CodeCodexCopilotDevinFactory
    Source control
    GitHubGitLabBitbucket
    Trackers
    JiraLinearPlane
    Artifacts
    JFrog ArtifactoryNexusContainer registries
    CI/CD & security
    JenkinsGitLab CISonarQubeSnyk
    Runtime & isolation
    Docker — isolated agent sandboxesReproducible · contained
    Deploy
    SaaSYour cloudAir-gappedAny model, any agent
    Governed by Reign

    Every change, provable.

    Reign is the governance layer beneath the factory: it enforces your policy at runtime and records what happened as tamper-evident, examiner-ready evidence. The audit becomes a question you've already answered.
    • ✓ Human approval gate — nothing reaches production without a person.
    • ✓ Runtime policy enforcement on every model and agent action.
    • ✓ Tamper-evident evidence — a sealed ledger for every change.
    • ✓ Deploy anywhere — SaaS, your cloud, or fully air-gapped.
    Evidence ledger · reign-x4
    ✓ ticket dispatched · worker claimed
    ✓ code + tests · all gates passed
    ✓ AI review · fixer cleared
    · merge to main → human approved
    — sealed to audit ledger —
    sha-256 9f2a…c71e · examiner-exportable
    Aligned to the rules that now govern AI
    OSFI E-23EU AI ActISO 42001NIST AI RMFFINOS AIGF
    Proof

    Not a demo. A factory we run ourselves.

    Our production apps already ship through Dark Factory — governed, audited and faster. We're now opening it to a small number of regulated customers through co-design pilots.
    21 Years
    Operating regulated infrastructure
    SOC 2 Type II
    Independently certified
    AWS Advanced
    & Validated MSP
    Sovereign
    Your data, models and agents stay yours
    Built on open foundations
    Member & contributor in the open standards behind governed AI — the Linux Foundation, FINOS, and the Agentic AI Foundation.
    The Linux Foundation — Silver Member
    FINOS — Member
    Agentic AI Foundation — Silver Member
    Co-design pilots · 2026

    See it run on your toolchain.

    We're opening a small number of co-design pilots with regulated enterprises. Start with a 30-minute working session.

    Request a 30-minute briefingTake the readiness assessment
    Limited 2026 cohort — by application only.