flowstack 1.0.0

AI Agent Platform SDK with built-in billing management

FlowStack SDK

From local agent to production in 5 minutes.

FlowStack is the simplest way to build and deploy AI agents. Write Python tools, test locally, deploy instantly. No infrastructure, no DevOps, no complexity.

🚀 Why FlowStack?

Think Vercel for AI agents. You focus on agent logic, we handle everything else.

• ✅ Write tools in Python - Natural @agent.tool decorator
• ✅ Deploy in one command - agent.deploy() and you're live
• ✅ Instant production API - Get HTTPS endpoints immediately
• ✅ Built-in memory - DataVault for persistent agent state
• ✅ Multi-provider support - Switch between OpenAI, Claude, Llama seamlessly

🎯 Perfect For

• 🥷 Indie developers - Build weekend projects that scale
• 🚀 Startups - Rapid prototyping without infrastructure overhead
• 🤖 Automation enthusiasts - Replace Zapier with custom AI logic

⚡ Quick Start

1. Install & Deploy in 5 Minutes

# Install
pip install flowstack

# Get your API key from flowstack.fun
export FLOWSTACK_API_KEY="fs_..."

2. Build Your First Agent

from flowstack import Agent

# Create agent with tools
agent = Agent("customer-helper")

@agent.tool
def lookup_order(order_id: str) -> dict:
    """Look up customer order status"""
    # Your business logic here
    return {"status": "shipped", "tracking": "UPS123456789"}

@agent.tool  
def update_address(order_id: str, new_address: str) -> str:
    """Update shipping address for an order"""
    # Your business logic here
    return f"Address updated for order {order_id}"

# Test locally
response = agent.chat("What's the status of order #12345?")
print(response)  # Agent calls lookup_order() automatically

3. Deploy to Production

# Deploy with one command
endpoint = agent.deploy()
print(f"🎉 Your agent is live at: {endpoint}")

# Now available as REST API:
# POST https://your-agent.flowstack.fun/chat
# {"message": "What's the status of order #12345?"}

🧠 Persistent Memory with DataVault

Agents remember context across conversations:

@agent.tool
def save_preference(user_id: str, preference: str) -> str:
    """Save user preference"""
    agent.vault.store('preferences', {
        'user_id': user_id, 
        'preference': preference
    })
    return "Preference saved!"

@agent.tool
def get_recommendations(user_id: str) -> list:
    """Get personalized recommendations"""
    prefs = agent.vault.query('preferences', {'user_id': user_id})
    # Use preferences to customize recommendations
    return ["item1", "item2", "item3"]

🔄 Multi-Provider Flexibility

Switch AI providers without changing code:

# Use managed Bedrock (included in pricing)
agent = Agent("my-agent", model="claude-3-sonnet")

# Or bring your own OpenAI key  
agent = Agent("my-agent", 
    provider="openai",
    model="gpt-4o", 
    byok={"api_key": "sk-your-key"}
)

# Same tools, same deployment, different AI provider

📊 Real-World Examples

Slack Customer Support Bot

agent = Agent("support-bot")

@agent.tool
def search_knowledge_base(query: str) -> str:
    # Search your docs/FAQ
    return search_results

@agent.tool  
def create_ticket(issue: str, priority: str) -> str:
    # Create Zendesk/Jira ticket
    return ticket_id

# Deploy and connect to Slack webhook
endpoint = agent.deploy()

E-commerce Order Assistant

agent = Agent("order-assistant")

@agent.tool
def check_inventory(product_id: str) -> dict:
    # Check your inventory system
    return {"in_stock": True, "quantity": 50}

@agent.tool
def process_return(order_id: str, reason: str) -> str:
    # Handle return logic
    return "Return processed"

# Deploy and embed in your website
endpoint = agent.deploy()

Content Creation Workflow

agent = Agent("content-creator")

@agent.tool
def research_topic(topic: str) -> str:
    # Research from multiple sources
    return research_summary

@agent.tool
def publish_to_cms(title: str, content: str) -> str:
    # Publish to WordPress/Contentful
    return "Published successfully"

# Deploy and trigger via webhook
endpoint = agent.deploy()

💰 Simple Pricing

Session-based pricing. One API call = one session, regardless of message count.

• Free: 25 sessions/month
• Hobbyist: $15/month, 200 sessions
• Starter: $50/month, 1,000 sessions
• Professional: $200/month, 5,000 sessions
• Enterprise: Custom pricing, 25,000+ sessions

📚 Learn More

• 📖 Documentation - Complete guides and examples
• 🚀 5-Minute Tutorial - Get started immediately
• 🧠 DataVault Guide - Persistent agent memory
• 🍳 Recipe Collection - Real-world examples

🛠️ Development

# Local development
git clone https://github.com/flowstack-fun/flowstack.git
cd flowstack
pip install -e .

# Run tests  
pytest tests/

# Build docs
mkdocs serve

🤝 Support

• 🐛 Issues: GitHub Issues
• 💬 Community: Discord
• 📧 Email: [e-mail]
• 🐦 Updates: @flowstack

---

Built for developers who ship fast. 🚢

Stop building infrastructure. Start building agents.