Describe it. Approve it. Own it.
Talk to your platform in plain language. The Createrun Agent builds real, reviewable applications on the same platform your developers use — and nothing ships without your approval.
Say what you need. Watch it take shape.
Describe a form, an approval process, a report. The Createrun Agent creates it as real C# code on the same platform your developers work on every day — not a throwaway prototype in a tool nobody owns. You watch the application take shape, ask for changes in the same conversation, and approve it when it's right. Even a big step — say, turning a flat request form into a multi-stage approval process — is one instruction, applied as one clean, reviewable change.
Build and evolve real applications — with the AI client your team already uses.
Build and evolve real Createrun applications from the MCP-aware AI client your team already uses. The Createrun Agent works under the same rules, the same permissions, and the same audit trail as your team. Supported operations are versioned in a capability matrix published with each release, marked live, partial, or planned (public matrix — Roadmap).
BuildScreens, fields, and the data behind them — created and changed in conversation.
DesignMulti-stage business processes with the right people approving at the right steps.
AnswerAsk questions of your data and turn the answers into reusable reports.
VerifyRehearse changes in an isolated test mode — never against production data.
ShipRelease when you approve; step back to the previous version in one click.
GovernManage access and review a complete record of who changed what, and when.
Your engineers can read the capability reference in the Developer Hub.
AI proposes. Createrun validates. People approve. The process engine executes.
The Createrun Agent never touches your systems directly. It proposes changes; a deterministic process engine — not the model — carries them out; and a human signs off before anything that matters takes effect. The AI's creativity goes into the proposal, never into the execution.
Reviewable batches, not scattered edits.
Related changes land together as one reviewable batch: you see exactly what will change before it does, and nothing takes effect until it is approved. Every change passes the same strict validation as work done by hand, and every action is attributable to who asked for it and who approved it. Releases can be stepped back to the previous version.
- Staged changes, one-click rollback — changes are staged before they land, and any release can be stepped back to the previous version in one click.
- A recycle bin for everything — anything the Agent removes goes to a recycle bin first. Nothing is silently destroyed; restore is always an option.
- An isolated test mode — the Agent rehearses in an isolated test environment. It never experiments on your production data.
- Attributable by design — every action traces back to who requested it and who approved it, end to end. Automated credential rotation is on the public roadmap.
The AI client is your choice. Bring your own LLM key, or run only the models your organization has approved. The platform runs on your infrastructure — including fully offline environments where no data ever leaves your network. Model costs stay visible and yours to control: your key, your bill, no markup hidden inside a per-seat price. The Createrun Agent is Createrun's own governed authoring surface; the same supported operations are open to external MCP-aware AI clients — including Claude Code, Codex, and Cursor — so your organization keeps the AI workflow it has already approved.
Watch the Agent build a real application.
We'll point the Createrun Agent at a sandbox instance and let it build a working CRM from a plain-language description — live, on a call, with you giving the instructions and granting the approvals.
Book a live build session