AI Features

Build AI-powered features with Vercel AI SDK and Agents

AI Features

Eden Stack includes a comprehensive AI suite built on Vercel AI SDK and Inngest AgentKit:

  • @eden/ai - Lightweight wrapper for chat and completions
  • @eden/agents - Advanced multi-agent systems with tool calling and durable execution

Simple Chat Completion

The @eden/ai package provides a unified interface for chat completions:

import { generateChatResponse } from "@eden/ai";
 
// Streaming response using default model (gpt-4o-mini)
const response = await generateChatResponse([
  { role: "user", content: "Hello, world!" }
]);
 
// Response is a streamText result from Vercel AI SDK
for await (const chunk of response.textStream) {
  process.stdout.write(chunk);
}

Required environment variable:

OPENAI_API_KEY=sk-...

AI Routes in Elysia

Expose AI functionality to your frontend with streaming routes:

// apps/api/src/routes/ai.ts
import { Elysia, t } from "elysia";
import { generateChatResponse } from "@eden/ai";
 
export const aiRoute = new Elysia({ prefix: "/ai" })
  .post("/chat", async ({ body }) => {
    const { messages } = body;
    return generateChatResponse(messages);
  }, {
    body: t.Object({
      messages: t.Array(t.Object({
        role: t.Union([t.Literal("user"), t.Literal("assistant")]),
        content: t.String()
      }))
    })
  });

Agent Networks

For complex tasks requiring reasoning, planning, or external data, use @eden/agents:

Planning Network

Breaks down complex goals into executable steps:

import { createPlanningNetwork, runPlanningAgent } from "@eden/agents";
 
const network = createPlanningNetwork();
const result = await runPlanningAgent({
  network,
  goal: "Create a marketing plan for the product launch",
});
 
console.log(result.output);

Research Network

Uses Exa's deep research API for comprehensive web search and analysis:

import { exaDeepResearch, exaSearch } from "@eden/agents/exa";
 
// Deep research - multi-step pipeline with citations
const research = await exaDeepResearch(
  "Latest trends in React server components"
);
console.log(research.content);   // Grounded report
console.log(research.sources);   // Citation URLs
 
// Quick search - for simpler queries
const quick = await exaSearch("React 19 release notes");

Additional environment variables for agents:

ANTHROPIC_API_KEY=sk-ant-...
EXA_API_KEY=...
TAVILY_API_KEY=tvly-...

Durable Agents

For long-running tasks that shouldn't fail on server restart, integrate with Inngest:

import { inngest } from "@eden/jobs";
import { durablePlanningFunction } from "@eden/agents";
 
// Trigger a durable agent as a background job
await inngest.send({
  name: "agents/planning.run",
  data: {
    goal: "Analyze competitor pricing strategies",
    model: "gpt-4o"
  }
});

Durable agents store state and execution history, allowing them to run reliably for minutes or hours.

Next Steps

Full documentation for Eden Stack users

This documentation is exclusively available to Eden Stack customers. Already purchased? Log in to access the full content.

Already a customer? Log in
View pricing details