Solutions
An internal app platform for apps your teams build with AI
An internal app platform is the shared foundation a company uses to build, secure, deploy and run the software its own employees use, such as approval workflows, request trackers, dashboards and admin panels. Diakrin is an internal app platform for apps built with AI coding agents such as Claude Code, Cursor and Codex: each app runs in the company's own AWS or Google Cloud account, and IT controls who builds, what each app can reach, who approves releases and what each app may cost.
Last reviewed
An internal app platform is the shared foundation a company uses to build and run the software its own staff use: approval workflows, request trackers, dashboards and admin panels. Diakrin is an internal app platform built for the way these apps are now written, by employees working with AI coding agents. The apps run in your own cloud account, and IT stays in control of who builds, what each app can reach, who approves releases and what each app may cost.
What is an internal app platform?
Most companies have dozens of small internal tools. Each one needs the same things: a sign-in that only admits the right employees, a database, a place for secrets, a build and deployment path, a way to approve changes, and someone who can see what it costs and what data it holds.
An internal app platform provides those parts once, so every new app does not rebuild them. Earlier internal tools platforms did this with a visual editor and connectors. With AI coding agents, writing the screens is no longer the slow part. The work that remains is turning a prototype into a company app that IT can trust. That is the part Diakrin handles.
How is an AI-era internal tools platform different?
Traditional internal tool builders ask builders to learn their editor and component model. Diakrin does not replace the editor. Builders describe the app they need and build it with Claude Code, Cursor, Codex or another MCP-capable tool. The agent connects to Diakrin with the builder's own permissions, reads the app's approved requirements and writes ordinary source code.
| Concern | Typical internal tool builder | Diakrin |
|---|---|---|
| Where the code is written | The vendor's visual editor | The builder's own AI coding tool |
| Where apps run | Often the vendor's cloud, or a self-hosted install | Your AWS account or Google Cloud project |
| Release control | Varies by plan | Every production release goes through an approved plan |
| Data access | Connectors configured in the builder | Named, scoped grants that IT approves for the app and for each employee |
| Cost control | Usually per seat | A monthly budget ceiling per app, checked before each deployment |
What should IT require from an internal app platform?
Before a platform hosts apps that touch company data, IT usually asks for the same controls. Diakrin documents each of them:
- Company sign-in. Employees open apps with your existing sign-in, and each app has its own audience. See Sign in to company apps.
- Separate access decisions. A successful sign-in does not grant access to any app or its data. See Manage access and security.
- Approved data access. A capability the app requests is not a capability it has. IT approves a named operation with resource and field limits. See Review capabilities and deployment.
- Independent release approval. Under your approval policy, an eligible reviewer who is not the requester approves the exact release plan. See Preview, approve and deploy.
- Budgets enforced before deployment. A deployment that would exceed an app's monthly ceiling is refused at admission.
- Evidence and audit. Builds, deployments and approvals keep their stages, receipts and approvers.
Where do the apps and their data live?
With BYOC (bring your own cloud), Diakrin runs the control plane and your cloud account runs everything else. Each app gets its own service, database, file storage and secrets inside a company cloud setup that IT creates from a reviewed template. On AWS, apps run on ECS on Fargate with Amazon RDS PostgreSQL and Amazon S3. On Google Cloud, they run on Cloud Run with Cloud SQL PostgreSQL and Cloud Storage.
Diakrin reaches your cloud only through access you create. On AWS that is one DiakrinAccess CloudFormation stack per account; on Google Cloud it is keyless workload identity in the project you choose. If you also need the console and API inside your network, the self-hosted edition runs the whole platform in your AWS account.
How it works with Diakrin
- IT prepares the company. Connect company SSO, invite people with the right roles and create a company cloud setup from a reviewed template, with a monthly budget ceiling. See Get started.
- A builder describes the app. Name it, explain what it should do, who will use it and what data it needs. Start with synthetic data. See Create your first app.
- The coding agent builds it. Claude Code, Cursor or Codex connects through MCP or the CLI with the builder's permissions and writes the code. See Connect your coding tool.
- Your cloud builds and checks it. The Diakrin runner builds an immutable image from an exact source commit and a pinned template, and scans it inside your account.
- IT reviews the plan. Diakrin checks the release plan against your rules: approved capabilities and data grants, the budget ceiling, the reviewed template and the approval policy. An eligible reviewer approves or denies it.
- The app goes live and stays observed. It counts as live only with fresh, server-verified evidence. IT and the builder track health, cost, data and activity, and can change the audience or stop the runtime.
What does it not do yet?
Scheduled jobs, queues and general internet access from apps are not offered on any cloud yet. Azure is coming soon. Some features differ by cloud; for example, the app data inventory and company-managed source are available on AWS and not yet on Google Cloud. The company cloud guide lists what each cloud provides today.
To see the path end to end, read How it works, or browse use cases for the apps teams build first.
Frequently asked questions
What is the difference between an internal app platform and a low-code internal tool builder?
A low-code builder gives you a visual editor and its own component model. An internal app platform covers the whole path to a running company app, including sign-in, access, source, builds, checks, approvals, deployment, budgets and audit. Diakrin does not include a visual editor. Builders describe the app and write it with the AI coding tool they already use.
Where do apps built on Diakrin run?
In your company cloud account. With BYOC, each app gets its own service, database, file storage and secrets in your AWS account or Google Cloud project. Azure is coming soon. With the self-hosted edition, the console and API also run in your AWS account.
Which coding tools work with Diakrin?
Claude Code, Cursor, Codex, OpenCode and other tools that support MCP over Streamable HTTP. A command-line tool, the Diakrin CLI, gives terminals and agents the same capabilities.
Who can build apps?
IT decides. Roles include Owner, IT administrator, Technical administrator, Builder and Member. Builders work within the apps assigned to them, and new app creation follows your company policy, which can require approval.
Does Diakrin see our business data?
Diakrin stores metadata such as app definitions, policies, approvals and cost accounting. App databases, files, secrets, builds and execution stay in your account. Its data view reports sizes and names, never row values, file contents or secret values.
What kinds of apps is it a good fit for?
Internal web apps and APIs, such as approval workflows, request trackers, operations dashboards and admin panels. Scheduled jobs, queues and general internet access from apps are not offered yet.
See Diakrin with your own cloud
Let your team build internal apps with the AI tools they already use, while IT keeps control of identity, data, approvals and spend.