A complete AI Assistant inside your .NET App

Three tools, not one per API

An App's whole API surface is far too large to keep in context. Only a compact search index is loaded eagerly, and a Model pays for an API's schema only when it actually uses it.

• api_search Finds APIs relevant to the user's intent
• api_describe Returns full schemas and workflow metadata
• api_call Invokes the typed Request DTO as the current user

[Tag("CoffeeShop")]
[Tool("the user wants to place an order",
    Safety = ToolSafety.Write,
    RequiresApproval = true)]
public class CreateCoffeeShopOrder : IPost { }

Consequential calls need a human

Reads execute immediately. Writes and destructive operations pause and render an editable form generated from the API's own schema — so the user sees exactly what will be sent, and can fix it before approving. No per-API Chat component to write.

Edit with AI

Describe the change in English. The model gets the current template, its data and any referenced partials, and Studio compiles what it returns immediately — with a repair pass if it doesn't compile the first time.

Add a Paid watermark when the outstanding balance is zero.

Move the totals into a bordered box and show tax on a separate row.

Reformat this as a 4×6 shipping label with no page margins.

AI edits update editor buffers rather than committing files, so experimentation stays reversible.

From template to inbox

Publishing validates the template and every fixture, generates the typed model, compiles it all and saves an immutable revision — rolling back if anything fails. Production rendering then needs nothing but typst and the published files.

public class InvoiceServices(IPdfRenderer pdf) : Service
{
    public async Task<object> Any(GetOrderInvoice request)
    {
        var order = await Db.LoadSingleByIdAsync<Order>(request.Id);
        return await pdf.PdfResultAsync(MapToInvoice(order),
            $"Invoice-{order.InvoiceNo}.pdf");
    }
}