For the HNG Internship, I built a modular text transformation chain using FastAPI microservices and fully integrated it with Telex.im’s AI coworker platform.
Workflow:
Input Agent: Receives text, coordinates calls.
Capitalizer Agent: Converts all text to uppercase.
Reverse Agent: Reverses word order.
Example:
Input: "hello world again"
Output: "AGAIN WORLD HELLO"
Deployment:
Each microservice is deployed independently on Render, communicating via HTTP endpoints.
The Input Agent orchestrates the chain and exposes a /transform-text endpoint.
Integration:
On Telex.im, I registered my input agent, described its workflow, and verified with test prompts.
(Side note: Still troubleshooting the category/publishing error, but setup is ready and endpoint works fine!)
Tech Used:
FastAPI
Uvicorn
httpx
Render for cloud hosting
Telex.im for agent collaboration
Challenges & Fixes:
Wrestling with JSON validation for category, I switched to defining tasks directly in the Telex UI—this is more reliable and makes agent setup seamless.
How To Try:
Send a POST to my endpoint with any sentence: get back uppercase and word-reversed result instantly!
Example:
json
{ "text": "hello world again" }
Returns:
json
{ "text": "AGAIN WORLD HELLO" }
Next Steps:
Once publishing is fixed, my agent will be live for anyone on Telex.im.
Follow my repository [your GitHub repo link] for source, docs, and updates.
Conclusion:
Microservice agent chaining is a scalable pattern for AI and automation.
Happy to assist with integration tips—ping me for help.
Top comments (0)