Operations Center Launching soon

An AI automation factory for enterprise teams.

Operations Center turns a plain-language request into a tested, security-reviewed and compliance-checked Ansible pull request. Self-hosted in your environment, governed by your engineers.

  • Reuse detection before any generation
  • Automated quality, security and compliance gates
  • Molecule and live AWX runtime evidence
  • Human approval before anything reaches production
Pipeline run #8421 all checks passed
  1. Reuse detectionreuse scored
  2. Blueprint builderstructure defined
  3. Quality gatestandards enforced
  4. Runtime + AWX QAruntime validated
  5. Hardeningcan block
  6. Compliancecan block
  7. Deploy — pull request deliveryhuman approval

Product status & availability

Launching soon

In active development. Join the waitlist for early access and release notes.

Guided demo

Walkthroughs are given live, on request, against a non-production environment.

Deployment

Self-hosted only. Your Git, your AWX, your secrets, your data boundary.

Capability descriptions on this page reflect the product's current implementation. Anything not yet released is labelled as such — nothing here is presented as production-ready before it is.

Seven-stage pipeline

Every automation passes through quality, security and compliance.

Seven stages run in order, each producing artifacts. Pull-request delivery is the result of the pipeline, not a stage inside it.

  1. Stage 0: Reuse detection

    Existing roles, playbooks and collections are searched and scored before anything is generated. Reuse is preferred over generation.

    • Six-factor similarity scoring across the automation catalog
    • Exact matches returned immediately, near matches extended rather than rebuilt
    • Match confidence reported to the requester

    Evidence

    2 reusable roles found

    similarity scored

  2. Stage 1: Blueprint builder

    A repository and role design is produced before code: structure, variables, tasks, handlers, tests, job template and a documentation plan.

    • Repository structure
    • Role and variable design
    • Handlers and task breakdown
    • Molecule scenarios and documentation plan
    • AWX job template

    Evidence

    Blueprint generated

    repository structure

  3. Stage 2: Quality gate

    Architecture review plus static analysis in one pass: yamllint and ansible-lint, naming, modularity and idempotency expectations.

    • ansible-lint and YAML validation
    • Architecture review against written standards
    • Naming and modularity checks
    • Idempotency expectations

    Evidence

    Lint passed

    2 warnings auto-fixed

  4. Stage 3: Runtime + AWX QA

    Molecule scenarios in disposable containers, then against real managed nodes over SSH and WinRM, then a live AWX job launch. A failure in any phase feeds an automated fix loop.

    • Molecule scenarios in Docker
    • Molecule against real managed nodes (SSH and WinRM)
    • Live AWX job execution, stdout captured
    • Test logs and job artifacts retained

    Evidence

    12 / 12 scenarios passed

    AWX job evidence attached

  5. Stage 4: Hardening

    Security review before a human ever sees the change. This stage can block delivery.

    • Secret detection
    • CVE and dependency review
    • CIS and risky-module checks
    • Secure defaults; secret-manager lookups where configured

    Evidence

    Security clean

    0 critical, 1 medium

  6. Stage 5: Compliance

    Governance gates and evidence retention. Runs in parallel with hardening, and can also block delivery.

    • Data anonymization checks
    • AI governance record and risk classification
    • Privacy and auditability review
    • Evidence retained against the change

    Evidence

    Compliance passed

    evidence retained

  7. Stage 6: Deploy — pull request delivery

    A Git-based pull request carrying the full evidence trail, opened for human review, with the work item updated and artifacts linked. This is the pipeline's result: nothing reaches production without peer review and merge.

    • Pull request opened in your Git platform
    • Full evidence attached to the change
    • Work item transitioned to peer review
    • No automatic production deployment

    Delivery

    PR opened

    awaiting review

Text alternative: seven stages run in order — reuse detection, blueprint builder, quality gate, runtime and AWX QA, hardening, compliance, and deploy. Hardening and compliance run in parallel and either can block. The pipeline's result is stage six: a pull request opened for human review.

What the product actually does.

Reuse first

Finds and adapts existing content before generating anything new, so estates converge instead of sprawling.

Built-in quality

Linting, architectural review and idempotency expectations are gates, not suggestions.

Real testing

Molecule scenarios and live AWX execution — evidence from a real runtime, not a model's assurance.

Secure by default

Secret detection, risky module policy and secure defaults applied before review, with Vault where configured.

Compliance ready

Policy results, AI governance records and retained artifacts, attached to the change that produced them.

Human approval

The pull request stays blocked until an engineer reads the evidence and merges it. No autonomous deployment.

Example journey

One request, end to end.

1 · The request

> Rotate local admin credentials on all RHEL 9 hosts in the EU estate, pull secrets from Vault, record evidence in ServiceNow.

2 · What the pipeline produced

  • 2 reusable roles found and extended
  • role + molecule scenario + README
  • ansible-lint clean, 2 warnings auto-fixed
  • AWX job evidence attached
  • no secrets in source, Vault lookup used

3 · The delivery

PR #184 opened, awaiting review. An engineer reads the diff and the evidence, then merges — or sends it back with comments. Production deployment stays a human decision.

Architecture & governance

Self-hosted, auditable, built on open components.

A request API and LLM orchestration layer drive agent services that write automation content; testing, security and compliance services validate it; Git and work management carry the result. Evidence and artifacts persist in your own data layer.

  • Runs inside your network — no automation content leaves it
  • Scoped tool access, least privilege, secrets via Vault
  • Full evidence retention for audit and change records
  • Extensible: add your own checks, policies and integrations

TODO — confirm the released component list against the product repository before publication.

Orchestration

FastAPI · Claude Agent SDK · agent workflow services

Automation & testing

Ansible · AWX 24 on k3d · Molecule · ansible-lint · yamllint

Delivery & data

Forgejo · Plane · MinIO · PostgreSQL · Harbor (optional) · n8n (optional)

Full architecture page →

Get early access to Operations Center.

Waitlist members get release notes and the first self-hosted evaluation builds. Prefer to see it first? Ask for a live walkthrough against a non-production environment.

Join the waitlist Request a demo

No newsletter, no tracking pixels. Your address is used to contact you about Operations Center only.