InvestEasy
Team consisting of Accenture full‑stack engineer Léo Chardon and Sahar backend developer Gianni Jouve, ESIEE Paris grads skilled in Java, Python, Angular, Spring, microservices, OpenAPI
YouTube Video
Project Description
AI Market Analyst targets solo analysts and founders who need quick market intelligence. It ingests a startup idea, runs live Mistral reasoning for summaries, positioning, profitability, and competitor scouting, then enriches the output with Qdrant-driven similar companies and optional n8n emailing.
Judges can follow the demo: submit an idea → view /analyze JSON in FastAPI docs → trigger /trigger-agent to watch the n8n workflow send curated competitor news.
Execution & Functionality – Full FastAPI backend with typed endpoints ( granular /analysis/*, /export/pdf) plus a working n8n automation (/trigger_agent). Live LLM + vector similarity calls prove end-to-end functionality.
AI & Agents Usage – Uses Mistral chat for reasoning, embeddings for vector search, and an n8n “agent” that curates RSS feeds and emails clients autonomously.
Scalability (One-Person Model) – Docker + Cloud Run deploy keeps ops simple; dataset sync script plus automated webhooks mean a single operator can run it.
Problem Clarity & Market Impact – Solves the pain of founders lacking quick competitive intel; delivers actionable insights (scores, profitability, similar startups) and a follow-up agent that pushes news to inboxes.
Demo & Product Narrative – Storyline: founder submits idea → backend generates VC-grade brief → optional agent emails daily intel; all visible through /docs + n8n UI.
Partner Tech Usage – Relies on Mistral API (reasoning + embeddings), Qdrant Cloud for similarity, and Google Cloud Run for hosting.
Autonomous agent logic: n8n workflow takes {idea, email, analysis} payload, fetches Google News RSS for competitors, formats personalized emails, and sends via SMTP. Failure modes: missing env vars (Mistral/Qdrant), empty Qdrant collection, or n8n webhook offline—handled with HTTP error responses and logs; next steps include persistent n8n storage and richer scoring feedback loops.
Technologies used – FastAPI, Pydantic, Uvicorn, Mistral AI API, Qdrant client/cloud, ReportLab, n8n, Google News RSS, Python dotenv, Docker, Google Cloud Run, Cloud Build, Google Cloud Shell, CSV dataset (“unicorns till sep 2022”).