Mosayc.ai

Inspiration

Our inspiration came from witnessing professionals struggle with AI research tools they can't trust. When making multi-million dollar investment decisions, these experts would start with AI research but ultimately discard it due to unverified data sources, black-box reasoning, and lack of transparency. We realized there's a massive opportunity to create an accountability layer that transforms AI from an interesting starting point into a trusted advisor for high-stakes financial decisions.

What it does

Mosayc provides expert-grade economic research you can trust by combining three essential elements missing from current AI solutions: verified data sources, expert-designed reasoning frameworks, and complete transparency. Our platform allows financial professionals to define research queries, then guides them through a structured process that ensures reliability at every step. Users can trace every data source, understand each reasoning step, and customize the methodology to match their organization's requirements. The result is AI-powered research that is beyond a Q&A.

How we built it

We are using a multi-modal, multi-agent infrastructure that uses CrewAI and uses collaboration to achieve a research task. We are using streamlit for UI (although we intent to have a more sophisticated UX with react), Gemini, GPT, Claude for research planning and execution, CrewAI for agents, Langtrace for observability, Elevenlabs for voice modality, Windsurf for IDE, MCP all to create a research output with curated logic on financial analysis.

Challenges we ran into

Python dependencies :P We tried vibe coding but realized sometimes its faster to just ignore all suggestions :) We also faced some challenges integrating UI and backend, some of the tools we used were not well documented. Apify didn't really work for us.