$npm i
Build AI agents as code
Write TypeScript directly in your React Native or Next.js app - agents deploy automatically with your code. Or build standalone workflows and deploy explicitly via CLI. No separate backends, no API mappings, no deployment friction. Ship AI features faster with zero infrastructure overhead.
1calljmp.ai.agent(
2 {
3 name: 'Support Assistant',
4 description: 'Answers customer questions using knowledge base',
5 },
6 async (input: { question: string; userId: string }) => {
7 // Generate embeddings with automatic retry
8 const embedding = await ai.retry(
9 { retries: 3, backoff: 'exponential' },
10 ai.phase('Embed Question', () => {
11 return generateEmbeddings(input.question);
12 })
13 );
14
15 // Search knowledge base
16 const articles = await ai.phase('Search Knowledge Base', () => {
17 return ai.vector.search(embedding, { limit: 5 });
18 });
19
20 // Generate personalized response
21 return ai.phase('Generate Answer', async () => {
22 const response = await ai.llm.generateText({
23 prompt: input.question,
24 context: articles,
25 userId: input.userId,
26 });
27
28 await ai.emit('answer_ready', { response, userId: input.userId });
29 return response;
30 });
31 }
32);
Embed & search
Transform customer question into vectors and search knowledge base. Uses ai.retry for fault tolerance.
Generate answer
LLM creates contextual response using retrieved articles. Each phase is traced for debugging.
Stream to app
Emit realtime events with ai.emit so your React app can show progress and partial results.
Observe and debug AI agents
Realtime insights into AI workflows with logs, and tracing tools. Monitor invocation latency, errors, and usage in the dashboard; stream realtime results to your apps. Trace multi-step agents end-to-end for quick issue resolution.
- Performance metrics
- Stream logs from AI operators to your dashboard for real-time insights into workflow execution.
- Evals
- Test agent and workflow outputs using rule-based and statistical evaluation methods to ensure high-quality responses and highlight improvements for your AI integrations.
- Tracing
- Agents emit traces for faster debugging and application performance monitoring, giving end-to-end visibility into multi-step executions and errors.
Secure, resilient AI for your apps
Calljmp's SDK empowers React Native and web developers with zero-config AI integrations. Built on edge-native BaaS, it ensures security, resilience, and seamless features like streaming chatbots - all in TypeScript.
Resilient AI workflows
Configure retry policies with exponential backoff and error handlers directly in your TypeScript workflows. Recover from transients like network flakes or model timeouts without crashes.
Secure AI invocations
Enable app attestation for mobile (React Native/Flutter) to verify devices and prevent tampering. For web, add bot protection and abuse detection - integrated with auth for per-user security.
Chatbot hooks
Build interactive AI chatbots using hooks like useChat for realtime streaming responses. Supports context persistence, tools, and agents.
Text generation
Generate text with useTextGeneration hook, supporting streaming for progressive UI updates. Integrate with RAG or agents.
Edge-native database
Built on Cloudflare D1 (SQLite) with row-level security, type-safe queries via Kysely, and realtime updates.
Realtime updates
Subscribe to database changes and AI events via WebSocket. Get live agent responses and workflow updates.
Cloud functions
Deploy TypeScript functions as edge workers for custom AI logic, wrappers, and integrations.
Trusted by developers
Accelerate AI-enabled app development with TypeScript - defined agents and workflows.
Simple, transparent pricing for apps
Start free, scale with predictable costs. AI invocations include primitives, agents, and chains - the backend (auth, database, storage, realtime) is bundled in every plan.
Free
$0/month
For prototyping AI agents. Start free, scale later. AI invocations include primitives, agents, and chains - backend free.
- 10,000 AI invocations / mo
- All SDKs (React Native, Flutter)
- Backend: auth, D1, R2, realtime
- Community support
Pro
$20/month
For production AI workflows with predictable scale and priority response times.
Everything in the Free, plus:
- 1,000,000 AI invocations / mo
- Unlimited agents & chains
- Additional usage $1 / million
- Priority support
Scale
$199/month
For enterprise AI automation. Custom limits, SLAs, and premium onboarding.
Everything in the Pro, plus:
- Custom usage & SLAs
- Premium support & onboarding
- Compliance & governance tools
- Strategic architecture reviews
From our blog
Insights, tutorials, and stories from the world of mobile development. Learn how to build better apps with less complexity.
Frequently asked questions
Find answers to common questions about Calljmp's features, pricing, and capabilities.