Jev in production › Model and agent routing

finance-agent (AmishKakka)

Jev replaces the LLM router in a personal multi-agent finance app, cutting routing latency from a full LLM call to about 100-400ms (author).

Open on GitHub ↗

100-400msmeasured against a baseline, as published by the source
Use
Model and agent routing
Industry
Finance
Form
Open-source tool
Stage
Beta
Listed
2026-09-29
Found via
discord
Repository
AmishKakka/finance-agent
Stars
2
Forks
0
Last push
2026-09-23
Language
Python
License
none stated

The README opens with

Takes a company ticker + user query and produces a focused investment research report by routing to specialized agents.

1. Supervisor (Jev) – Decides which agents are needed using calibrated probabilities 2. Selected agents run in parallel (LangGraph) 3. Controller – One call to Gemini synthesizes all findings into a clean report

Only the relevant agents are used, keeping it fast and cheap.

Agent What it does --------------------- NewsAgent Recent news & events (Tavily) FinancialStmtAgent Income statement, balance sheet, cash flow, earnings (yfinance) OutlookAgent Analyst recommendations, price targets, revenue & growth estimates (yfinance) SectorAgent Sector / industry level news & outlook

Badge

For the project's own README, linking back here:

Listed in Jev in production

Also used for model and agent routing