AI Native Platform
The cloud for
AI agents.
Define agents, harnesses, and logic in config. Ship with one command. Run indefinitely in the cloud. One API. Any model. Spawn one, or thousands.
{
"name": "support",
"model": "gpt-5.5",
"tools": [
"check_order",
"send_email",
"search_docs"
],
"skills": [
"billing",
"refunds"
],
"prompt": "You are a support..."
}
platformagent
tools
skills
sandbox
shared fs
models
My subscription did not renew.
Charge failed. I retried it and sent confirmation #A3F7.
· designed to work with
· the problem
Your agent is not the hard part. The runtime around it is.
The moment an agent does real work, your team inherits sandboxes, streaming, memory, queues, retries, tools, routing, and observability. Build that first, and you are shipping infrastructure instead of product.
Runtime plumbing, integration loops, and millions of tokens before the agent is production-safe.
Define the agent harness and focus on its functional logic, not on the runtime around it.
· meet polpo
The AI-native platform that handles the hard parts.
A composable agent harness plus AI Gateway, giving your agents the runtime, tools, sandboxes, model routing, and feedback loops they need.
Composable agents
Define what your agent knows, does, and can access. Model, tools, skills, and system prompt, all in a single portable config.
Built-in sandbox integration
Agents run in isolated sandboxes to access shell, filesystem, browser, and network. Provisioned for you, not by you.
Chat with any model
Streaming OpenAI-compatible chat endpoint. Tools, vision, structured outputs across every provider.
Long-running work, not scripts
Dispatch tasks with dependencies, checkpoints, retries, and multi-step execution for production agents.
Agents on your clock
Agents run for minutes or hours. Trigger them on demand, from webhooks, or on schedule.
Keep outputs accountable
Judge outputs, retry failures, and keep production agents accountable.
· how it works
Agent config becomes a production API.
Keep the agent definition close to your code. Polpo turns it into a production API your app can call from anywhere.
Define the agent
Model, prompt, tools, and skills live in a small config your team can review and version.
{"name": "coder","model": "openai/gpt-5.5","systemPrompt": "You are...","tools": ["bash", "read", "..."],"skills": ["testing", "..."],// rest of the agent config}
Deploy the runtime
Polpo provisions the hosted runtime, sandboxes, memory, and endpoint behind the agent.
$ polpo deploy✓ agents 3✓ skills 4✓ memory mounted✓ endpoint /v1live in 4.2s
Call it from your app
Use one OpenAI-compatible API for chat sessions, tasks, and agent work.
await fetch("/v1/chat/completions", {method: "POST",body: JSON.stringify({agent: "coder",stream: true,messages: [...]}),});
· how it fits
Add agents to your stack with one API.
Keep your product as it is: call Polpo from your app, backend, or workflow. We run the agent sessions, scheduled work, sandboxes, and model routing behind it.
· coding agents in action
From prompt to production.
In seconds, any coding agent can generate the full harness: agents, tools, skills, memory, deploy wiring, and instantly get a live API.
Use cases
What teams build on Polpo
App builders
Build AI-powered apps with agents as the backend. Ship features, not infrastructure.
like Lovable, Bolt, Replit Agent
AI employees
Agents with identity, memory, and skills that work as autonomous team members. 24/7, no burnout.
Zero-human companies, digital workforce
Voice agents
Plug voice agents into your call center or assistant. Real tools, real actions, real-time.
Support bots, sales assistants
Agent control panels
Build dashboards to manage your agents. Monitor runs, inspect logs, control execution — all via API.
Internal tools, admin panels
Autonomous workflows
Multi-agent missions that run on schedule. Triage tickets, generate reports, monitor systems.
DevOps, support, data pipelines
Portable agents
Spin up disposable agents for batch jobs, evaluations, or as extensions of your own coding agent. CLI or API, no infra to babysit.
Batch jobs · Evaluations · Coding-agent extensions
· faqs
Questions, answered.
The short version. For deeper answers, the docs are exhaustive.
read the docsPolpo is the agent layer for your app. It runs conversational and long-running agents in isolated sandboxes, gives them persistent memory, tools, skills, and vault, and exposes a unified chat API so your app can call them over HTTP.
· start building