guuey.blog
Tutorials

Ship your first hosted agent on Guuey

From blank page to a live, streaming agent on its own sandboxed pod — the full Guuey Studio walkthrough, plus the CLI path if you'd rather type.

Read as .md

You bring the instructions. Guuey brings the pod, the stream, and the playground. This tutorial takes you from a blank form to a deployed agent your users can actually talk to — no SDK to import, no event handlers to write, no code at all unless you want some.

What you’ll build

A live hosted agent. It runs continuously on its own gVisor-isolated pod, streams replies over Server-Sent Events directly from that pod, and gets its conversation history persisted automatically — you never build a chat store. With Generative UI on (it’s the default), your agent replies with interactive screens, not just text. That comes from mcp.ggui.ai, the default MCP server, which speaks ggui — an open, MCP-native generative-UI protocol. Guuey and ggui compose, but neither requires the other.

Prerequisites

  • A Guuey account — sign up at guuey.com.
  • A browser. That’s the whole toolchain.

Create your first agent

  1. Sign in at studio.guuey.com. Email and password always work; Google and GitHub sign-in are offered where available. One account carries you across Guuey.
  2. Start a new agent. On your first visit, Guuey Studio points you straight at the create form — click Create your first agent.
  3. Write the instructions. This is the heart of your agent: who it is, what it does, how it behaves. Not sure where to start? Click a starter template chip — Trip planner, Meeting-prep assistant, Customer FAQ answerer, Study coach, Social media copywriter, or Meal planner — and edit from there. Or open Draft it for me, type one sentence, and Studio expands it into full instructions.
  4. Name it. Templates propose a name you can shuffle or replace.
  5. Pick a model. Three plain-language tiers: Thinking for the deepest reasoning, Balanced (recommended), or Fast for the snappiest, lowest-cost replies. An Advanced section exposes the underlying provider, model, and framework if you want them.
  6. Choose superpowers. Generative UI is on by default. You can also add your own MCP server by URL under Advanced.
  7. Click “Create my agent.” Studio creates the agent and deploys it in one step, then drops you into its playground. Watch the status chip in the header walk from Starting… to Live.

The playground is a phone-framed chat wired to your actually-deployed agent — the same live endpoint your users will reach. Give it a poke. Then open Settings to tweak the instructions, model, or superpowers and hit Deploy changes.

Prefer code?

Studio is one of two ways in. Developers can scaffold a code-mode agent — bring Claude Agent SDK, OpenAI Agents SDK, or Google ADK code — run it locally, and deploy it with the CLI:

npx @guuey/create-agentic-app my-agent
cd my-agent
pnpm install
pnpm dev      # run locally with hot reload

guuey login
guuey deploy  # hosted on Guuey

After deploying, guuey test "hello" sends a test message and prints the response, and guuey logs --follow live-tails your runtime logs. The CLI ships on npm as @guuey/cli; you’ll need your Guuey account to log in and deploy.

What just happened

When the chip hit Live, Guuey booted your agent on a long-running, gVisor-sandboxed pod — a much stronger isolation boundary than an ordinary shared container. Long-running means exactly that: nothing sleeps between conversations, so the first message of the day is as fast as the last. Your chat streamed straight from the pod with no relay in between, and every turn was persisted for you. And the whole agent is one declarative definition — a system prompt, a model, MCP servers — validated the same way by guuey dev, guuey deploy, and the hosted runtime. One contract everywhere. No wobble in the contract; all the wobble in the UI.

Where to go next

Ship the prompt. The interface figures out the buttons.