Compare · camelRun by camelAI
Vercel eve alternative
Your AI agent as an API you call from any host.
camelRun by camelAI is a hosted runtime for durable AI agents. Keep each agent's configuration in agent.yaml next to your code.
const run = await agent.run("Summarize ticket 123", { user: "alice" });const run = await agent.run("Summarize ticket 123", { user: "alice" });Migrating from eve
Move eve agents to camelRun
Your prompt stays a Markdown file and your tools stay TypeScript. Here is where the parts of an eve agent go.
- agent.tsmodel: anthropic/claude-sonnet-5-5 (agent.yaml)
- instructions.mdsystemPromptFile: prompts/support.md (agent.yaml)
- tools/mcpServers: [{ name: app, url: … }] (agent.yaml)
- channels/Console › Channels (console)
- schedules/agent.schedule({ text, everySeconds }) (your code)
- subagentsdelegate: { agents: [...] } (agent.yaml)
- approvalsneedsApproval: true (your tool)
After the move
Your eve agent as agent.yaml, next to your Next.js app
agent.yaml holds what agent.ts did and points at your instructions as a Markdown prompt. Your app serves the tools, and camelrun deploy puts the agent live from your laptop or CI.
The CLI and agent.yaml- acme-help/
- ├── app/
- │ ├── api/agent/route.ts
- │ ├── mcp/route.ts, from eve's tools/
- │ └── page.tsx
- ├── prompts/
- │ └── support.md, from eve's instructions.md
- ├── agent.yaml, from eve's agent.ts
- └── package.json
key: support
systemPromptFile: prompts/support.md
builtins: [web_search, ask_user]
mcpServers:
- name: app
url: https://app.example.com/mcp
auth: { type: bearer, token: "${APP_MCP_TOKEN}" }
agents:
- key: supportnpm install -g @camelai/camelruncamelrun login # an API key from the console's API tokens pagecamelrun init support # writes agent.yamlcamelrun deploy # the definition, and the agents it listsNext.js and the AI SDK
Keep your Next.js app and useChat on Vercel
Building Vercel agents with Next.js and the AI SDK? Keep your pages, routes and useChat where they are. Each customer gets an agent of their own in camelRun, made from the support definition you deployed, and replies stream from camelRun straight to the browser, so a long answer never holds a function open.
You're right: 3 of your 12 seats belong to people who left in September. Remove them under Settings
// app/api/agent/route.ts (Next.js App Router)
import { createAgentHandler } from "@camelai/run/server";
export const POST = createAgentHandler({
// CAMELAI_API_KEY from the environment, or apiKey: "…"
authorize: async request => {
const user = await getUser(request); // your session check
return user ? { userId: user.id, name: user.name } : null; // null: 401
},
agent: { definition: "def_…" },
});Connect your AI agent to Slack, Discord, Telegram and GitHub
Add a channel in the console and pick the agent's definition. Every Slack thread it's @mentioned in, DM, Telegram chat, pull request or issue gets an agent of its own, and webhooks from services like Sentry and Linear can start one too.
How channels workAgents in your account. Your application creates them with the SDK or POST /v1/agents. Learn more
| Name | Type | Model | Status | Session expires |
|---|---|---|---|---|
| support | general | anthropic/ | Running | Never |
| Slack: U07HM2QX4 | channel | anthropic/ | Asleep | Never |
| GitHub: acme/help#1187: Fix seat count on invoices | channel | anthropic/ | Running | Never |
| Telegram: @jonas_k | channel | anthropic/ | Asleep | Never |
| Discord: @ravi | channel | anthropic/ | Asleep | Never |
| Webhook: sentry-4821 | channel | anthropic/ | Asleep | Never |
Vercel eve or camelRun: which fits your project
- Your team ships everything on Vercel and wants each agent to be its own project directory.
- You want Vercel Sandbox, Workflows and AI Gateway underneath your agents.
- You want to call agents from the app you already run, in TypeScript or Python, with tools that stay in your own code.
- You want to pay only while agents work: $0.01 an agent-hour in a run, prepaid, and idle or sleeping agents cost only storage.
- You want to run the whole runtime on your own servers, or bring your own model keys and model servers.
Both are open source: eve under Apache-2.0, camelRun's runtime under the AGPL with MIT SDKs and CLI.
eve details from vercel.com/eve and github.com/vercel/eve, checked October 22, 2026.
Questions about moving from Vercel eve
What is Vercel eve?
The eve agent framework is Vercel's open-source framework for durable AI agents, in beta since June 2026. An eve agent is a directory of Markdown instructions and skills, TypeScript tools, channels and schedules, which eve deploys to Vercel.
What's the difference between Vercel eve and camelRun?
Vercel eve is a framework: an agent is a project directory, tools included, that you deploy to Vercel. camelRun by camelAI is a hosted runtime: each agent lives under a key you pick, its tools stay in the app you already run, and your code calls it from any host in TypeScript or Python.
Can I run durable AI agents outside Vercel?
Yes. camelRun runs your agents at run.camelai.com or on your own servers with one container and Postgres. Your agents keep going through deploys and crashes, and a tool call is never sent twice.
Can I keep my Next.js app and the AI SDK on Vercel?
Yes. Add one route with createAgentHandler, point useChat at it with AgentRuntimeChatTransport, and serve your tools with serveTools. Using Vercel AI Gateway? Add its key on the console's Models & keys page.
Can my camelRun agent answer in Slack and Discord like an eve agent?
Yes. Connect Slack, Discord, Telegram, GitHub or a signed webhook in the console, with no channel code, and each conversation gets an agent of its own.
How do I move an eve agent to camelRun?
Point agent.yaml's systemPromptFile at your instructions.md, set the model, and serve your tools from your app as an MCP server that agent.yaml names. Then run camelrun deploy, add your channels in the console, and move your schedules to agent.schedule().
What does a durable AI agent cost on camelRun?
camelRun is prepaid, with no subscription, no seats and no per-agent fee: $0.01 for each hour an agent spends in a run, $0.10 per GB-month of storage, $0 for tokens on your own provider keys, the provider's price with no markup on tokens on ours, and idle and sleeping agents cost only storage.
Your AI agent as an API
npm install -g @camelai/camelruncamelrun logincamelrun init supportcamelrun deployRead https://run.camelai.com/SKILL.md and set up camelRun in this project. Ask me what the agent should do.
Paste into Claude Code, Codex, Cursor or any coding agent. It asks what your agent should do, then builds it.