Open-source · MIT license

Citadel-grade research throughput. Without the Citadel-sized headcount.

Six AI workflows that turn hours of analyst grunt work into minutes — thematic screens, investment memos, IC prep, catalyst maps, cross-asset checks, and scenario analysis. Institutional-grade output, every time.

Coming soon: the Engine. Numbers you can trust, analysis you can score, deliverables you can hand to your PM.

Works with Cursor, Claude Code, OpenClaw, and Cowork.

The largest funds already built this internally.

Citadel, Balyasny, and Point72 have deployed AI research assistants that compress weeks of analyst work into minutes. Their edge isn't the model — it's the structured workflows, data plumbing, and quality governance underneath.

You need the same throughput. You don't have the same headcount.

Six workflows. Hours of work, compressed into minutes.

Each workflow encodes institutional best practices into a repeatable decision framework. Free, open-source, and ready to use in your IDE today.

Thematic Screening

Go from a macro thesis to a ranked list of investable names in minutes — with the non-obvious second and third-order plays most screens miss.

See example output

Investment Memo

Produce an IC-ready memo that argues a position, not just describes a company. Variant perception, key drivers, valuation, risks — done in one pass.

See example output

CIO/PM Q&A Prep

Know the 25 hardest questions before you walk into the room. Pre-built answers, supporting evidence, and clear red lines for each.

See example output

Catalyst Mapping

See every event that could move your position — with asymmetry analysis that tells you which catalysts are worth trading around.

See example output

Cross-Asset Signals

Find out whether rates, credit, FX, commodities, and vol agree with your thesis — or are quietly contradicting it.

See example output

Scenario Analysis

Know your upside, downside, and exactly where the thesis breaks. Probability-weighted returns with sensitivity tables and decision triggers.

See example output
Coming Q2 2026

Stop estimating. Start computing.

The free skills give you structured analysis with estimated figures. The Engine is a paid upgrade that gives you real math — computed sensitivity grids, Monte Carlo simulations, and quality-scored output you can hand directly to your PM.

Free Skills
Engine (paid)
Output
Markdown document
Structured JSON + formatted PDF
Numbers
LLM estimates (flagged)
Computed from valuation models
Sensitivity
3×3 table (estimated)
N×N grid (calculated) + Monte Carlo
Quality
Unchecked
Scored against explicit criteria
Data
Training data only
Live data via MCP connectors
Integration
Copy-paste
API, calendar export, JSON for downstream tools

Every output scored against 162 institutional quality criteria. The same evaluation framework that maintains a 96.3% pass rate internally now powers the Engine's user-facing quality scores.

Trust the numbers.

Sensitivity grids calculated from valuation models, not guessed by an LLM. Monte Carlo across 1,000 iterations. You see exactly where your thesis breaks.

Know where your analysis is weak.

Every output scored against institutional quality criteria. You see the per-section breakdown and specific suggestions for what to strengthen before IC.

IC-ready deliverables, not raw text.

Formatted PDF, structured JSON, and clean markdown. Hand it to your PM, feed it into your models, or pipe it to another tool via API.

The Engine launches in Q2 2026 as a Pro subscription. Join the waitlist for early access and launch pricing.

For funds that want this embedded in their research process.

We work directly with mid-sized investment firms to integrate custom AI workflows into daily research operations — tailored to your process, your data, and your team. Your analysts cover more names. Your research process becomes repeatable.

Phase 1

Diagnostic & Workshop

We run the system on your coverage universe, map your research workflow, identify bottlenecks, and deliver a concrete implementation roadmap. You keep everything — whether or not you go further.

Phase 2

Custom Implementation

Production research flows tailored to your process, wired to your data, embedded in your daily tools, with a firm-specific quality evaluation suite and ops runbook.

It works today with public data. It becomes your edge with proprietary data.

Start getting value immediately with the free skills and public data. Connect your licensed research, internal KPIs, or portfolio data through the Engine — and the same workflows produce materially sharper output.

Surface Mode

Public filings and AI estimates. Every estimate flagged for verification. Structured, useful, ready to build on.

Enhanced Mode

Your licensed data, internal metrics, and proprietary signals. Institutional-grade, differentiated output that reflects what you actually know.

Built on the Model Context Protocol (MCP). FRED macro data is active and free. Paid providers (FactSet, S&P Global, PitchBook, Morningstar, etc.) plug in via your existing API keys. Your data stays local — nothing leaves your environment unless you send it to the Engine API.

Questions

Who is this for?

Investment professionals who use AI coding tools (Cursor, Claude Code, OpenClaw) for research. The free skills work for anyone. The Engine and institutional services are designed for analysts, emerging managers, small funds, and family offices.

Do I need an API key for the free skills?

No. The skills run on whatever LLM your IDE provides. No additional API keys, no account required.

Are the numbers in the output real?

In Surface Mode, market data and estimates come from the LLM’s training data and are flagged with [estimate]. Connect MCP data sources or use the Engine for computed, verified figures.

What’s the difference between the free skills and the Engine?

The free skills produce structured markdown with estimated numbers. The Engine computes real sensitivity grids, runs Monte Carlo simulations, scores output quality, and returns structured JSON + formatted PDF. The skills are the methodology. The Engine is the math.

How is this different from ChatGPT / Claude?

ChatGPT and Claude are general-purpose models. Investment Intelligence is a structured decision architecture built specifically for institutional investment analysis. The difference is the same as between a blank spreadsheet and a fully built financial model.

Is my data safe?

The free skills run entirely locally in your IDE. The Engine API processes inputs server-side — we do not train on your data, and all data is encrypted in transit and at rest.