AIchimist

• 

  Default Interface (Login, Signup and settings included).

• 

  Used 3 AI models (Learning boost: ON. Requires 3 or more models )

• 

  Used 2 AI models (Memory boost: ON. One AI answers other keeps context on Vertical approach)

• 

  Respond from "GemiSeek" (mix of Gemini and DeepSeek).

Video is LOONG please check screenshots!

Please don't forget to check screenshots too! If video is corrupted!

Inspiration | _ ALchimist's who's making from iron and other metals, GOLD! _

What it does | _ Mixing AI for more efficient for better and smarter results! Fewer hallucinations, Memory Boost and MIXED AI self learning Boost by having multiple opinions and self-critiqe of inputs and outputs for better future respond. _

How we built it | _ React Native + Expo frontend with physics-based UI (AI Google studio). Bun + Hono + tRPC backend. Direct API integration to 4 AI models. W&B Weave logging via GraphQL. _

Challenges we ran into | _ API balance issues. Model deprecations. tRPC path conflicts. Designing real hallucination detection algorithms. _

Accomplishments that we're proud of | _ ** Built a new AI paradigm. AIchimist! ** _

What we learned | _ Sucrificing your sleep WORTH the mission! _

What's next for AIchimist | _ Add more models. Visual metrics dashboard. Community-shared "recipes". Enterprise API. Become the Docker for AI ensembles. _

link

Built With

• aigooglestudio
• claudeai
• deepseekai
• geminiai
• mistralai
• react
• rork
• trcp
• ts
• w&b
• w&bsdk
• w&bweave
• windsurf