Self-hosted
Run VizRepo in your own cloud, with the LLM you already approved
Install VizRepo in your AWS or Azure account and connect it to Azure OpenAI or Amazon Bedrock. Your code stays in your network. We are not a vendor processing your data. We are software you run.
Why this makes approval easy
Most of a vendor review exists because a third party will hold or process your source code. Self-hosted removes that question.
- No third party receives your source code. Not us, and no AI provider you have not already approved.
- No new data processor to assess. VizRepo runs under your own cloud account, policies and access controls.
- AI traffic goes to the model your company already cleared, under the contract and retention terms you already have.
- Your security team can inspect the deployment, the network rules and the logs themselves.
- Procurement becomes a software purchase, not a data-processing agreement.
How it runs
| Part | Where it runs | Notes |
|---|---|---|
| VizRepo API and scanner | Your AWS or Azure account | Containers on ECS, EKS, AKS, App Service or a VM |
| Web app | Your account | Behind your SSO proxy or private network if you prefer |
| Database | Your MongoDB or a managed equivalent | Holds the generated documentation only |
| AI model | Azure OpenAI, Amazon Bedrock, or your own provider key | Requests never pass through VizRepo's servers |
| Source code | Read from your Git server inside your network | Cloned for the scan, deleted after, never stored |
What your teams get
- Every endpoint documented from its code: purpose, business rules, steps, data read and written, external services, errors.
- Docs that update on every merged pull request, with a plain-language change entry.
- A data-access map showing which endpoint touches which table.
- Sync to your internal Confluence, SharePoint, Azure DevOps Wiki, GitHub Enterprise Wiki or Notion.
- An Ask box that answers questions from the verified documentation, with sources.
- Works with GitHub Enterprise, GitLab self-managed, Azure DevOps, Bitbucket and any Git server reachable from the deployment.
For the project manager
- New engineers understand the system in days, without booking the one person who knows it.
- When someone leaves, what they knew about the code stays.
- You can check how a feature really works without pulling an engineer off their task.
- Reviews and handovers start from a current description of the system, not a page from last year.
- No writing effort from the team. The documentation is generated and maintained for them.
Rollout
- Week 1: a call to confirm your cloud, Git server and approved model. We provide the deployment package and configuration.
- Week 1 to 2: your platform team deploys it with our support. First repository scanned the same day.
- Pilot: two or three services, measured on onboarding time and how often engineers are interrupted with how-does-this-work questions.
- After the pilot: roll out to the remaining services. Updates are delivered as new container versions you apply on your schedule.
Not ready for self-hosting?
Two lighter options exist. Use our EU cloud, where source code is never stored. Or use our cloud with your own AI provider key, so the AI traffic runs under your contract. You can move to self-hosted later without losing your documentation.
Questions
- Does anything leave our network?
- Only what you allow. VizRepo talks to your Git server, your database and the AI endpoint you configure. There is no call home with code or documentation.
- Which models are supported?
- Azure OpenAI and Amazon Bedrock (including Claude on Bedrock), as well as direct Anthropic or OpenAI keys. Documentation quality is best with Claude models.
- Do you need access to our environment?
- No. Your team runs the deployment. Support happens over calls and whatever logs you choose to share.
- How is it licensed?
- An annual license sized to the number of repositories and engineers. AI usage is billed by your own cloud provider, so there are no credits.
- Can we start with a pilot?
- Yes. A paid pilot on two or three services is the usual first step. Write to info@vizrepo.com or use Talk to Sales.
- What does our platform team need to provide?
- A container runtime, a MongoDB-compatible database, network access to your Git server and to your AI endpoint, and a read-only Git token.
Try it on your own repository
Connect a repo, run one scan, and read what each endpoint does. Starter is free. Pro has a 14-day free trial.