Joel Tan@joelTXW

Documents are uploaded, parsed and stored as markdown files where the agent will have access to in order to gain relevant information to come up with a response for the user. Uses tools like grep to look for patterns and read_range to read a range of characters in the document it chooses to read through. It will also have the ability to break down the task given to smaller tasks and give it to a subagent to do which will report back with its own insights to the agent that called it. Using all these information, the main agent will come up with a response and pass it on to a judge agent. The judge agent will then do its own research and verify if the response is a satisfactory answer to show the user. If it is, it will return that response to the frontend. If not, it will return the prompt back to the main agent with the rating and its reasoning for the rating so the main agent can improve upon it. This loop happens for maximum 5 times to prevent infinite recursion. If the max retries is hit, the best response in the 5 tries will be returned.

Grab your ideas, spin up compute, and start building what’s next! Start with $100 in AMD Developer Cloud credits and begin building right away. Get access to hands-on tutorials, workshops, and curated learning resources to help you build faster and go further. To unlock credits and join the hackathon, sign up for the AMD AI Developer Program. 🤝 Go solo or team up — ship something real. 💰 $21,500+ total prize pool & 1 × AMD Radeon™ AI PRO R9700 GPU up for grabs. 📍 On-site Venue (May 9-10): MindsDB SF AI Collective 3154 17th St, San Francisco, California, USA 📍On-site participation is by invitation only. Travel and accommodation expenses will not be covered. 🧑‍💻 Developers, builders, and innovators - bring your boldest ideas and let’s turn them into real AI systems on AMD.