Generated document with code

Inspiration

Our inspiration came with the idea of making an agent that could not only answer a prompt but making a plan and execute it, going along the user and taking responsibility into the researching, writing and programming tasks, as it adapts to the users way of working.

What it does

The Architect is planned to transform s complex task into actionable workflows, it receives and input from the user and generates a plan, if the user accepts the plan, then the agent starts with it until finishing, the agent deliberates the work into different modules depending on the requirements of the task, being able to do research and write detailed documents, and even programming new modules itself if its required, allowing The Architect to expand its capabilities.

How we built it

We built The Architect as a modular, multi-agent system. It is built it in python and integrates LLM based reasoning with our modular architecture and cloud infrastructure, we also used initially Groq APIs, and later on, Gemini's.

Challenges we ran into

One of our biggest challenges was designing a system where all our modules could work together without losing the context of the initial request. Another challenge was determining a complete and efficient presentation of the work of each module than needed it, we needed that our modules could present their work on an organized and clear way, for example the writing and programming module, which needed to receive, organize, analyze and present the work in an organized way. We also had problems when communicating as a team, as each of us worked separately on a module, we had to wait for others to finish a module and upload it to Github, to join them together and test them. Also we have problems because most of the time the APIs were saturated and we could not realize tests.

Accomplishments that we're proud of

We learned how to use an API to connect our application with an AI model and turn our idea into a functional tool. We learned that creating an AI application is not only about getting the model to generate responses, but is crucial for it to understand and process the information sent by the user.

What we learned

We learned how to use an API to connect our application with an AI model and turn our idea into a functional tool. We learned that creating an AI application is not only about getting the model to generate responses, but is crucial for it to understand and process the information sent by the user.

What's next for The Architect

Our next step is to continue improving The Architect, making it more autonomous and capable of learning from its previous work, also we want for the agent to be able top learn about the user's way of writing to adapt to it but also for write as it would, to have a closer approach to the work, we want The Architect to develop from an agent that gives you to a more efficient and productive responses that transform an idea to execution.