AI Agent Orchestrator ProjectsAI Agent Orchestrator ProjectsIn Numun.io (http://Numun.io), workflows are first designed and validated in Agent Studio as structured, multi-step graphs—made up of agents, tools, conditionals, loops, and outputs. Once a workflow is complete, it doesn’t live in isolation.
By toggling the workflow on in Agent+ Mode, users can execute that same workflow directly from the AI chat interface.
Instead of receiving a simple text response, the AI runs the full workflow behind the scenes and returns structured run results, including tool outputs, decisions taken, and final outcomes.
The result is an AI experience that goes beyond conversation, turning chat into a live interface for executing and inspecting real autonomous workflows.
In this example, Python code is generated and executed via Fly.io (http://Fly.io), returning a structured summary of results. I built a production Insurance AI Agent (https://github.com/Emarhnuel/Insurance-AI-agent.git) with FastAPI + CrewAI with 2 autonomous crews (3-agent Onboarding, 1-agent RAG), custom premium calculators, policy analysis tools, RAG knowledge base (7 PDFs), Mem0 memory, OpenTelemetry monitoring. Full K8s deployment (GKE), CI/CD pipeline, health checks, rolling updates.
Transform complex workflows into intelligent multi-agent systems. Check the GitHub link below
Stack: Python, CrewAI, FastAPI, Kubernetes, Docker, OpenAI, Gemini AI Builder + Agent Studio is a system for designing, testing, and deploying AI agents and autonomous workflows as structured, production-ready graphs, not fragile prompts.
At the core is Agent Studio, where workflows are defined as explicit graphs made up of agents, tools, conditionals, loops, routers, and outputs.
The AI Builder sits on top of this system and generates complete workflows automatically. Instead of writing code or wiring nodes manually, users describe the goal—and the AI Builder constructs a valid workflow graph.
Every workflow can be tested using a test harness, ensuring agents execute correctly before deployment.
This project demonstrates how AI agents can be treated like real software systems—versioned, validated, and orchestrated—instead of one-off prompt chains. 
