Skip to content
Back to projects

01 / Context

The problem

Long-running report generation needed to be independently deployable and observable, with every quantitative claim grounded in official disclosures rather than LLM-generated numbers.

02 / Implementation

What I built

Extracted report synthesis into its own FastAPI and LangGraph service, deployed independently as Synthesis, computing Sharpe, Sortino, drawdown, and holding-overlap metrics deterministically from ingested AMC factsheets, with tiered subscription access.

Current status
Deployed
Evidence path
2 public links

03 / Current outcome

What exists today

A live, independently deployed product at synthesis.fundersai.co.in with its own subscription tiers, decoupled from the core FundersAI application.

Boundary

What this page does not claim

Research-only; deterministic metrics are grounded in official disclosures, but it does not provide personalized investment advice or buy/sell/hold recommendations.

Continue exploring

Full archive