Run Pipecat on MIOSA

~15 min TypeScript Python

What you’re building: Install the open-source voice pipeline from speech-to-text to text-to-speech.

Primitives you’ll use: Sandbox, Redis

Agent prompt

Start with your coding agent

A few quick questions, then copy a prompt for any coding agent.

Who is it for?

Pick one to start. The rest adapts to it.

For my company Internal. One organization, a workspace per team, SSO. No billing. Build a product For customers. A workspace per customer, isolation, metering and billing.

What you're building

an open-source framework for real-time voice and multimodal agents Modelled on Pipecat, by Daily (open source).

Pipecat builds a pipeline of frame processors: audio in, speech-to-text, an LLM, text-to-speech and audio out, over a WebRTC or websocket transport.

Primitives you'll use: Sandbox · Redis

A pipeline for STT, LLM and TTS
Real-time transports (WebRTC, websocket)
Tool calling and function handlers
Open source and provider-agnostic

You install and run the real Pipecat on MIOSA. Read the official docs first: docs.pipecat.ai/getting-started/quickstart. Every install command below comes from them; if the docs and this guide disagree, the docs win. Pipecat is open source Python framework.

What you need on MIOSA

Each row is one thing to create before you start. The number matches the step that uses it.

Organization and API key Scopes every call; a workspace key is all a worker needs. miosa api-key create app-key --preset agent Docs
Sandbox The isolated Linux workspace the agent writes code and runs commands in. miosa create app-box --template nextjs --wait Docs
Redis Managed cache and session store, injected as REDIS_URL. miosa api POST /databases -d '{"name":"app-cache","engine":"redis"}' Docs
A workspace per customer Isolates each customer’s machines, deployments, and data as they onboard. miosa workspace create customer-1 Docs
Branding and white-label Your name and slug on previews, deployments, and the desktop; customers never see MIOSA. miosa org Docs
Usage metering and bill-to Usage per customer, and which account pays for new machines. miosa org bill Docs

Architecture

How it runs on MIOSA

Where it runs -> a MIOSA sandbox (an isolated Linux workspace)

State and files -> Session state belongs in managed Redis, not on the sandbox disk.

Model -> your own provider key, never a MIOSA platform key

Its web port 7860 -> a MIOSA preview URL (`miosa preview create`)

Your customers -> one workspace each, isolated from each other

Auth and tenancy

Your organization is the platform; customers never get a MIOSA account, they sign into YOUR product. One workspace per customer, created as they onboard, isolates their machines, runs and data.

miosa workspace create pipecat-customer-1

Tag every machine and run with the customer id in `metadata`, and never query across customers. Meter usage per customer with GET /api/v1/usage and set who pays with PUT /api/v1/bill-to (/docs/platform/usage-and-billing).

Install Pipecat

Run these inside the sandbox (for example `miosa exec pipecat-box -- bash -lc '<command>'`), in order. The commands are from the official docs.

curl -LsSf https://astral.sh/uv/install.sh | sh

uv tool install "pipecat-ai[cli]"

pipecat init

# or, by hand: uv init my-pipecat-app && cd my-pipecat-app && uv add pipecat-ai

Check: the install finishes without errors; the next sections configure and start it.

Configure

Add the provider extras you use (`uv add "pipecat-ai[option,...]"`) and put your own speech-to-text, LLM and text-to-speech keys in `.env`. Telephony and WebRTC transports are external providers you bring.

Model: Anthropic (Claude). Calls use my own provider key (ANTHROPIC_API_KEY); MIOSA platform keys are never used.

export ANTHROPIC_API_KEY="..."   # set it in the process environment, or the Secrets API; never write it into files you commit

Run it

Start it and prove it works.

uv run bot.py

Check: the tool starts without errors and responds.

Persistence and access

Session state belongs in managed Redis, not on the sandbox disk.

It listens on port 7860. Run it as a background process so it outlives your shell, then open a preview:

miosa preview create pipecat-box 7860 --name web

Check: the preview URL answers, and still answers after the machine pauses and resumes (previews wake it on request).

Make it a product

One bot process per customer workspace (or a pool with a session router); session state in Redis keyed by call id.

Limits and costs

Pin the version you tested and update deliberately; these tools change quickly.

Give the tool only the credentials it needs, as secrets, not baked into files.

Prototype: keep it to one machine at the default size, skip replicas and custom hostnames you do not need, and delete everything when you are done.

Acceptance checks

Pipecat runs on a MIOSA sandbox and answers a real task.

State survives a pause and resume.

Each customer has its own workspace and its own instance; nothing is shared between them.

Everything it created can be deleted with nothing left running.