Back to Features
Build AI Agents
Create agents with your preferred framework. Deploy to the cloud. Use them from ChatGPT, Claude, or any AI assistant.
Pick Your Framework
LangGraph
Complex multi-step workflows with state management
Strands
Simple conversational agents with built-in memory
CrewAI
Multi-agent collaboration for complex tasks
OpenAI-compatible
Works with any OpenAI-compatible API
What Your Agents Can Do
40+ AI Models
Claude, Nova, Llama, Mistral - all through AWS Bedrock. Switch models anytime.
Use From Anywhere
Access your agents from ChatGPT, Claude Desktop, Gemini, or any MCP-compatible assistant.
Deploy in Minutes
One-click deployment to AgentCore Runtime. No infrastructure to manage.
Built-in Memory
Agents remember context across conversations. Event memory and semantic memory included.
Code Interpreter
Let agents write and execute Python code in secure sandboxes.
How It Works
- 1.Choose a framework and configure your agent (name, model, system prompt)
- 2.Connect knowledge bases and tools your agent needs
- 3.Click deploy - your agent is live in minutes
- 4.Use it from the web, API, or any MCP-compatible AI assistant