European-first · Cloud-agnostic · Self-hosted

Deploy AI agents on
infrastructure you control

Write an agent.py, run stackagent deploy, get a production agent with memory, tools, and streaming — on your own Kubernetes cluster.

Book a Demo See How It Works 
# agent.py
from stackagent import Agent, tool

@tool(name="lookup_order", description="Look up order by ID",
      parameters={"type": "object", "properties": {"order_id": {"type": "string"}}})
async def lookup_order(order_id: str) -> str:
    return f"Order {order_id}: shipped, arriving tomorrow"

@Agent(name="support", tools=["lookup_order", "http_get"], memory=True)
async def support(ctx):
    """You are a customer support agent. Help customers with their orders."""
$ stackagent deploy
✓ Deployed support-agent (v1, 1 replica)
$ stackagent invoke --agent support "Where is my order ORD-123?"
⚙ lookup_order({"order_id": "ORD-123"})
Your order ORD-123 has shipped and is arriving tomorrow!

Everything you need to run agents in production

No vendor lock-in. No data leaving your cluster. Just deploy and go.

psychology

Memory & Sessions

PostgreSQL + pgvector for conversation history and semantic search. Redis-backed sessions. Agents remember context.

build

Built-in Tools

HTTP client, sandboxed code interpreter, and custom @tool decorators. Add any capability in Python.

bolt

Streaming Responses

Real-time SSE streaming with token usage tracking. See tool calls and results as they happen.

widgets

Per-Agent Isolation

Each agent runs in its own container. Deploy, scale, and version independently. Gateway routes automatically.

shield

Sovereign by Design

Runs on any Kubernetes — STACKIT, OVHcloud, on-prem. Your data never leaves your infrastructure. GDPR-ready.

hub

Any LLM Provider

Connect to any OpenAI-compatible endpoint. Works with European and sovereign LLM providers:

Mistral (France) · STACKIT AI (Germany) · OVHcloud AI Endpoints (France) · Aleph Alpha (Germany) · DeepL LLM (Germany) · Self-hosted vLLM / TGI

From code to production in minutes

No Dockerfiles. No Kubernetes YAML. Developers focus on agent logic.

1

Install the platform

Your ops team runs this once on any Kubernetes cluster.

helm install stackagent oci://ghcr.io/baseflare/stackagent
2

Define your agent

Write skills in markdown. Add tools in Python. That's it.

stackagent init my-agent
3

Deploy

One command builds, pushes, and deploys your agent to the platform.

stackagent deploy
4

Invoke

Talk to your agent via CLI, API, or integrate into your application.

stackagent invoke --agent my-agent "Hello!"

Runs entirely in your cluster

No data leaves your infrastructure. Every component runs in your Kubernetes namespace.

StackAgent Architecture

Deploy sovereign AI agents today

Available for Kubernetes clusters on STACKIT, OVHcloud, and on-premise. Get a personalised walkthrough of the platform.

Book a Demo

Or email us at hello@baseflare.co.uk