Features
Everything below is already built into the platform, now in active development ahead of early access — two surfaces, one engine, and a corpus that spans public filings, internal data, and your own documents.
An analyst you can watch think
Build a structured analysis or just ask in chat — either way, the same multi-agent engine plans, delegates, and executes in the open.
Analysis builder
Configure a company deep-dive, comparison, or valuation with the scope and data sources you choose.
Live agent progress
Stream the Manager's plan, the task DAG, and every Worker tool call in real time over server-sent events.
Results dashboard
Charts, ratio sets, tables, and narrative — each artifact linked back to the tool run and sources that produced it.
Conversational chat
Ask follow-ups in plain language. Chat is backed by the same tools and corpus as the dashboard, with the same citations.
Answers with receipts
Hybrid retrieval over your entire corpus — filings, transcripts, news, and uploads — with citations you can open and check.
Hybrid search
BM25 full-text and vector search fused with reciprocal-rank fusion, plus optional reranking for higher precision.
Document grids
Run one question across up to dozens of documents at once and get a grid of answers with a citation in every cell.
Citation viewer
Click any claim to see the exact document chunk or line item behind it — provenance is stored, not inferred.
Company search
Find entities across SEC EDGAR and market data, then pull filings, facts, and prices into an analysis in one step.
A quant workbench in the same product
Derivatives pricing and risk tooling that research platforms simply don't ship.
Options pricing
Black-Scholes with full Greeks and binomial-tree pricing for European and American options.
Monte-Carlo simulation
Path simulation under GBM, Heston stochastic volatility, and SABR — with configurable paths and horizons.
Risk analytics
Historical and parametric VaR and CVaR on portfolios and single names.
3D volatility surfaces
Interactive implied-volatility surfaces rendered in 3D, straight from market option chains.
From screening to allocation
Systematic idea generation and portfolio construction, grounded in point-in-time data.
Rule-based screener
Screen universes on fundamentals, ratios, and price behavior with composable rules.
Backtests
Point-in-time portfolio backtests so results reflect what was knowable at the time.
Efficient frontier
Mean-variance optimization and holdings analytics across your watchlist or portfolio.
Watchlist alerts
Track companies you care about and surface in-app alerts on watchlist activity.
Work products, not screenshots
Analyses export to files your team can keep editing.
Excel models
Driver-based forecasts that flow into a live DCF — real formulas, ready for your own assumptions.
Word primers & briefs
Company primers, SWOT analyses, and news briefs generated as editable Word documents.
Share links
Share a completed analysis with a signed link — no login required for the reader.
Charts & candlesticks
Price history with candlestick and volume charts, ratio visualizations, and custom charts from the agent's own charting tool.
Your environment, your rules
GenXFintel is software you run — not a SaaS you send your data to.
Self-hosted
Deploy with Docker in your own infrastructure. Filings, internal data, and questions stay inside your perimeter.
Provider-agnostic LLMs
Point it at any OpenAI-compatible endpoint — Azure OpenAI, OpenRouter, or a local vLLM server.
Per-run cost tracking
Token usage and external API calls are metered per analysis, with enforced budget limits.
Extensible by design
Tools, data sources, and aggregation strategies are swappable interfaces — add one by writing one class and registering it.