ALM for safety-critical projects

One ALM over the tools you already use

Your requirements live in one tool, code in another, tests and documents in a third. ALM OS links them, shows what a change affects, and freezes everything into signed baselines an assessor can rely on — without moving your data out of the tools your teams already work in.

Built for railway safety work:
  • EN 50126
  • EN 50128
  • EN 50129
  • EN 50716
  • CLC/TS 50701
ALM OS home page for a requirements engineer: links that need review, coverage and recently changed requirements ALM OS home page for a requirements engineer: links that need review, coverage and recently changed requirements

Works with

  • Jira
  • GitLab
  • GitHub
  • Redmine
  • Azure DevOps
  • Confluence
  • SharePoint
  • DOORS
  • Polarion
  • Codebeamer
  • Gerrit
  • Jenkins
  • JFrog
  • ReqManTool

and test results from your CI (JUnit XML, Robot Framework) · All integrations

Evidence should not be a spreadsheet project

In a safety-critical project every requirement has to be traced to design, code and tests, every change has to be assessed, and every release has to be shown to the assessor exactly as it was. When that information is spread over five tools, it ends up in spreadsheets built by hand the week before the audit.

Traceability across tools

Links from a requirement in ReqManTool or DOORS to a merge request in GitLab, a story in Jira and a test result from CI — in one place, with one coverage matrix.

Change you can see

When either end of a link changes, the link turns to Needs review, with what changed, who changed it and when. Nothing silently stays green.

Baselines as evidence

Freeze a release across every tool at once. The baseline is stored in git, sealed with a signature and signed off by independent reviewers, verifiers and approvers.

Reports for the assessor

An evidence report per baseline: tools and versions, items with content hashes, links and their state, coverage and integrity checksums.

How it works

  1. Connect your tools

    A guided wizard connects each tool through its plugin. Items stay where they are; ALM OS keeps references and fingerprints, never a copy to keep in sync.

  2. Link and review

    Link items across tools in one dialog. ALM OS watches both ends and asks the owner to review a link when something changes.

  3. Freeze, sign, report

    Check that every tool answers, freeze the baseline, collect independent signatures and hand the evidence report to the assessor.

Start from the tools you have

Jira with no requirements tool, a team that lives in GitLab, Azure DevOps with SharePoint, an on-premise railway supplier: see how ALM OS fills in the rest of the ALM, or pick your own tools area by area.

  • A team that lives in GitLab

    No managed requirements, no requirement-to-merge-request trace an assessor can follow, and no record of which tests passed for a release.

    Learn more →
  • Azure DevOps and SharePoint

    Nothing ties a SharePoint document to the work items and tests it covers, and nothing freezes Azure DevOps and SharePoint together for a release.

    Learn more →
  • An on-premise railway supplier: Redmine, Jenkins and JFrog

    Requirements, the hazard log and code reviews with independence by SIL are kept by hand, and nothing shows which requirements, reviews, test results and binaries make up a delivered release.

    Learn more →
  • Requirements stay in DOORS or Polarion

    The requirements tool cannot see the issues, merge requests and test results, so the trace is a spreadsheet rebuilt for every release.

    Learn more →

Designed around the people who do the work

Home shows what each person should do today: a requirements engineer sees the links to review, a safety manager the releases waiting for sign-off, an assessor the signed releases and their evidence.

Needs review, with what changed

The exact words that changed, the version before and after, and a Still valid… dialog that records who confirmed the link and why.

All features →
The Needs review page of ALM OS: links to SRS-002 that need review, with the changed words highlighted The Needs review page of ALM OS: links to SRS-002 that need review, with the changed words highlighted
Screenshot of ALM OS 0.1 with demo data.

Sign-off with independence built in

Whoever froze a baseline cannot review, verify or approve it, and one person holds one role. Every signature is checked again each time the page is opened.

All features →
A release page of ALM OS: baseline R1.0 frozen, waiting for review, verification and approval A release page of ALM OS: baseline R1.0 frozen, waiting for review, verification and approval
Screenshot of ALM OS 0.1 with demo data.

AI agents propose. People decide.

Connect Claude Code, Claude Desktop, Copilot in VS Code, Gemini CLI, ChatGPT or Cursor to ALM OS over MCP. The agent reads your requirements, links and baselines as you and turns what you ask for into a proposal. Nothing is written until a person accepts it — and a safety project can require that person to be someone else.

AI agents in ALM OS
  • A guided set-up: one script per computer, for Windows, macOS or Linux
  • Proposals accepted in whole or in part, or rejected with a reason the agent reads
  • Accepted changes recorded as AI-assisted, in the audit trail and in the tools’ history
  • No AI model inside ALM OS: your agent brings its own

On your premises. Air-gapped if you need it.

ALM OS runs on your own server, in one Docker container or on Proxmox. Licences are signed files checked offline, there is no telemetry, and your kernel data is a git repository you own. A lapsed licence never locks evidence away.

Security and compliance
  • On your premises
  • Offline licences
  • No telemetry
  • Your data stays yours

See it on your own project

We are looking for teams who want to try ALM OS on a real railway project. Every plugin can be tried free for 30 days.