TanStack Start Integration

Use createIntentFormServerFn to run the engine inside a TanStack Start server function. The function executes on the server; the client calls it like a normal async function.

Install

pnpm add @intentform/server @tanstack/react-start

Server function

app/server/intent.ts

import { createIntentFormServerFn } from '@intentform/server'
import { createIntentForm } from '@intentform/core'
import { openaiProvider } from '@intentform/provider-openai'
import { myModels } from '../models'

const engine = createIntentForm({
  provider: openaiProvider({ apiKey: process.env.OPENAI_API_KEY! }),
  models: myModels,
})

export const resolveIntent = createIntentFormServerFn(engine)

Client component

app/routes/index.tsx

import { IntentForm } from '@intentform/react'
import { createClientIntentForm } from '@intentform/client'
import { resolveIntent } from '../server/intent'
import { myModels } from '../models'

const engine = createClientIntentForm({
  serverFn: resolveIntent,
  models: myModels,
})

export default function IndexPage() {
  return <IntentForm engine={engine} onSubmit={(v) => console.log(v)} />
}

The server function runs in the server bundle. No fetch configuration or CORS needed — TanStack Start handles RPC serialization.

See Server-Side Architecture for a full comparison of patterns.