Skip to content

FlowStack SDK

From Local Agent to Production in 5 Minutes

Build your AI agent locally. Deploy with one command. Get a production endpoint instantly.

  • 5 Minutes to Production

    Skip weeks of infrastructure setup. Build locally, deploy instantly.

    result = agent.deploy()  # That's it

  • Built-in Persistence

    DataVault gives every agent MongoDB-backed storage with zero configuration.

    vault.store('users', data)

  • Simple Billing

    Pay per session (API call), not per token. No surprise bills.

    Free: 25 sessions/month | Hobbyist: $15/200 sessions

  • Auto-Scaling Infrastructure

    We handle Lambda, containers, databases. You focus on agent logic.

    From 0 to 1000s of requests automatically.

The Problem

You can build an AI agent locally in 10 minutes.
Deploying it to production takes 10 days of DevOps hell:

  • ❌ Setting up AWS Lambda functions
  • ❌ Configuring Docker containers
  • ❌ Managing databases and persistence
  • ❌ Setting up API gateways and load balancers
  • ❌ Handling authentication and security
  • ❌ Managing scaling and monitoring

The FlowStack Solution

from flowstack import Agent

# 1. Build your agent (YAML or code)
agent = Agent.from_yaml("agent.yaml", api_key="fs_...")
# Or define in code with instructions
agent = Agent(
    name="customer-helper",
    api_key="fs_...",
    instructions="You are a helpful support agent"
)

# 2. Test locally (works on your machine)
response = agent.chat("What's the status of order 12345?")
print(response)

# 3. Deploy to production (one command)
result = agent.deploy()
print(f"Deployed: {result['namespace']}")

# 4. Your agent is now live and accessible via API

That's it. Your agent is now running in production with:

✅ HTTPS API endpoint - Ready for production traffic
✅ Automatic scaling - From 0 to thousands of requests
✅ Built-in persistence - DataVault for storing data
✅ Security & isolation - Each agent runs securely
✅ Monitoring & logs - Track usage and debug issues

Who This Is For

Indie Developers & Hackers

You're building side projects, prototypes, or internal tools. Pain point: You can hack something locally, but deploying sucks.

Weekend Project

Build a Slack bot, Discord assistant, or automation tool. Deploy Sunday night, use Monday morning.

Startups Exploring AI

You don't want to spend weeks on infrastructure when testing ideas. Value: FlowStack = shortcut to shipping production-ready agents.

MVP in Days

Test your AI product idea with real users. Iterate on features, not infrastructure.

Automation Enthusiasts

You're already using Zapier or n8n. Value: Agents + persistence + real runtime → more powerful than simple automation.

Zapier Killer

Build stateful workflows that remember context between runs. Handle complex decision trees.

What We Handle (So You Don't Have To)

  • Infrastructure

    Lambda functions, Docker containers, auto-scaling, load balancing

  • Persistence

    MongoDB-backed DataVault with namespaced, isolated storage

  • AI Access

    Managed Bedrock, OpenAI, Anthropic with simple provider switching

  • Security

    API authentication, request isolation, rate limiting, monitoring

  • Monitoring

    Usage tracking, error handling, debugging tools, real-time logs

  • Billing

    Transparent session-based pricing, usage tracking, no hidden fees

Ready to Start?

The Vision

FlowStack is the Vercel for AI agents. You build the logic, we handle everything else to make it production-ready. Skip the infrastructure grind — focus on what makes your agent unique.

Get Started →