Manat.ai | Devpost

Manat.ai Logo
UI 1
s

Inspiration Our journey began from a place of personal struggle, often encountering the bottleneck of validating numerous business ideas brewing in our minds. The scenario of scribbling down a groundbreaking startup idea in Apple notes is all too familiar, followed by the tedious chore of toggling between tabs to research its potential. This routine not only hindered our momentum but also clouded our judgment with information overload. It was this friction that birthed the idea of Manat.ai - a tool envisaged to streamline the initial validation of startup ideas using Language Model Machines (LLMs), serving as a preliminary sieve to filter through the feasibility of our potential startup ideas.

What We Learned The project was a rich learning ground. We delved into the art of analyzing business ideas, identifying key data points for an initial assessment. The event also brought us together as a team, teaching us the essence of collaboration. We explored various cool technologies including Langchain and OpenAI, and harnessed LLMs to fuel our application, marking our first stride into integrating such advanced AI into a project.

How We Built Manat.ai Utilizing the Flowise platform, we orchestrated LLM flows comprising nodes and chains from OpenAI and Flowchain. This structure allowed us to incorporate conversational agents and AutoGPT for smarter workflow designs. The backend was engineered using Redis, Flask, and Express to interface with APIs, handle Python analytics, and manage data storage. On the frontend, we employed Next.js and JavaScript, crafting a user-friendly interface to interact with our system.

Challenges Faced Our first time building an LLM application was a blend of excitement and hurdles. Designing chains and creating intelligent agents capable of web access and task execution posed a steep learning curve. The modular nature of our work - different teams focusing on LLM workflows, Python scripts for analytics, and others on integrating these into a cohesive application, brought about the challenge of assembling these disparate pieces into a unified, functioning system. Each challenge, however, was a stepping stone, propelling us closer to our vision of simplifying startup ideation through Manat.ai.