AI-Native Private Investment Firm
Judgment, compounded by software.

General Intelligence Capital

General Intelligence Capital deploys capital across public equities and private credit using proprietary intelligence infrastructure. AI agents that source, analyse, debate, and monitor investments — continuously improving through market feedback.

Two Strategies. One Operating System.

The infrastructure that identifies a mispriced equity is the same infrastructure that underwrites a credit facility. Running both on one platform creates compounding information advantages that siloed firms cannot replicate.

ATLAS — Public Markets
Specialised agents debate macro, rates, commodities, sectors, and single stocks across four layers daily. Every recommendation is scored against real outcomes. The system improves its own prompts through autoresearch and learns which agents to trust through Darwinian weight selection.
VALIS — Private Credit Secondaries
VALIS is an AI-native origination and underwriting engine for private credit secondaries. It identifies opportunities before they reach the market, analyses them at institutional depth, and monitors the entire portfolio continuously.
Origination
VALIS scans SEC EDGAR, DOL Form 5500, IRS 990 filings, Companies House records, and CAFR disclosures in real time. It detects signals of distress, fund restructuring, and motivated sellers — covenant pressure, management turnover, refinancing walls, LP liquidity needs — weeks before opportunities surface through traditional broker channels. 59 distress signals detected in a single day. 1,175 active pipeline companies under continuous surveillance.
Underwriting
Every opportunity passes through a multi-agent IC process. Cash flow agents, collateral agents, covenant agents, comparable transaction agents, and management quality agents debate each deal independently. An adversarial agent stress-tests every assumption. The system produces a full IC memo, bid recommendation, and risk register in minutes — work that traditionally takes a team of analysts weeks.
Portfolio Intelligence
Once capital is deployed, VALIS monitors the portfolio continuously. It tracks borrower health through filing updates, news sentiment, sector stress indicators, and payment patterns. Early warning signals surface automatically — deteriorating cash flow coverage, sector headwinds, management changes — giving the investment team time to act before problems compound.

GENERAL INTELLIGENCE ALL SEASONS

Different markets demand different intelligence.

We don't train one set of agents and hope they work everywhere. We train separate cohorts on distinct market regimes, then let a meta-layer decide which regime we're in based on which agents are getting things right.

Bull
Low Volatility Growth Agents trained on steady, low-volatility markets. They learn patience, concentration, and momentum.
Crisis
High Volatility Stress Agents trained on market dislocations. They learn capital preservation, hedging, and speed of exit.
Tightening
Rate Hiking Cycle Agents trained on rising rate environments. They learn macro sensitivity, sector rotation, and duration management.
Recovery
Post-Crash Rebound Agents trained on post-crisis rebounds. They learn to deploy capital aggressively when others are still defensive.
Euphoria
Speculative Mania Agents trained on momentum-driven markets. They learn positioning, flow, and when fundamentals disconnect from price.

A meta-layer called JANUS sits above all cohorts and algorithmically weights them by recent accuracy. The weight shift between cohorts is the regime signal. We didn't build a regime detector. It emerged.

Four Innovations. One Architecture.

The firm's entire investment process is software. Every day the system runs is another training iteration.

Autoresearch Inspired by Karpathy. Agent prompts are the weights. Sharpe ratio is the loss function. Each trading day is one training iteration. The worst-performing agent gets its prompt rewritten by the system. If performance improves, the change survives. If not, it reverts.
Darwinian Selection Every agent carries a weight between floor and ceiling. Top performers are amplified. Bottom performers are silenced. The system independently discovered its own portfolio manager was its weakest component — before we diagnosed the same issue manually.
Agent Spawning The system detects recurring knowledge gaps in its own debates. When the same blind spot appears across multiple days, it autonomously creates a new specialist agent. The spawned agent enters at neutral weight and must earn its place through selection — or go extinct.
JANUS Meta-Layer Multiple trained cohorts produce different recommendations. JANUS sits above all cohorts and weights them by recent accuracy. The weight differential between cohorts is an emergent regime detector.

GIC OS

We don't use AI as a tool. We built the firm's entire investment process as software.

GIC OS is the operating system underneath both strategies. It originates, underwrites, debates, executes, and monitors — continuously, across public and private markets. Every decision flows through the same infrastructure: the same data ingestion, the same adversarial review, the same audit trail.

The result is an investment process that is faster, cheaper, and more disciplined than what headcount alone can achieve — and one that improves with every cycle.

How We Invest

Every decision flows through the same infrastructure. Same data ingestion, same adversarial review, same audit trail.

Source
Continuously scans public and private markets for opportunities across the capital stack.
Analyse
Specialised agents run diligence, build models, generate IC-ready materials. Every opportunity stress-tested.
Debate
Agents argue for and against. The adversarial agent's only job is to kill the trade. What survives is higher conviction.
Decide
The Investment Committee reviews agent output alongside human judgment. Clear criteria. Documented decisions.
Execute
Capital deployed as equity, hybrid, or credit depending on where the best risk-adjusted return sits.
Monitor
Continuous telemetry, intelligent alerts, exit planning. The system never stops watching.

Milestones

2024
GIC founded. Investment Committee assembled. VALIS built solo in 8 weeks.
Early 2025
First deal pipeline. $20M senior secured LOI signed. VALIS origination engine live.
January 2026
ATLAS built. 25-agent swarm architecture. Four-layer debate system. Autoresearch operational.
March 2026
TLAS launched publicly. 700K+ views. 650+ GitHub stars. Three live portfolios deployed. All Seasons regime training underway.

Leadership

Chris Worsey

Founder & CEO

2x exited entrepreneur and investor. EB-1 visa holder for extraordinary ability in business.

Built version 1 of ATLAS and VALIS solo.

Matt Sidari

Head of Investments

Leads deal execution across the private credit strategy.

Investment Committee

Guido van Hauwermeiren

Committee Member

Vice Chair at Société Générale.

Josep Julia

Committee Member

Founder, Deva Capital.

Private credit and distressed investing.

Ross Eldridge

Committee Member

20+ years of private credit and private equity experience across $35B+ in deal volume. Built and led a de novo healthcare credit platform with zero realized credit losses.

Mounir Guen

Committee Member

One of the most connected fundraisers in global private equity, decades of experience in investor relations and fund structuring

Why General Intelligence Capital

The firm is the technology.
Most funds treat AI as a productivity layer. We designed the investment process as software from day one. The technology doesn't assist decisions — it structures how they get made.
Information compounds across both books.
A signal that moves our equity positioning also informs our credit underwriting. A private deal that reveals sector stress reshapes our macro view. Siloed firms cannot do this.
The cost structure is the edge.
AI agents running continuously cost less than a single junior analyst at a traditional platform. That asymmetry widens every year as models improve and compute gets cheaper.

The moat is the training data. The trained prompts are proprietary — evolutionary products of hundreds of days of market feedback. A competitor starting today is hundreds of iterations behind. That gap widens every day.

For Founders & Owners

We are not a passive investor. Our process is fast, transparent, and aligned.
Faster diligence
days, not weeks — faster reporting, smarter decisions, and higher output per FTE.
Sharper insight
AI-built market maps, customer calls, KPI benchmarking
Operating lift
Telemetry, automation, competitive tracking
Exit readiness
Acquirers identified early materials prepared well ahead of time

For LPs, investors, and potential team members.

In public markets, expect systematic alpha generation with full transparency into the agent decision-making process — every trade documented from first signal through adversarial review to execution.

In private credit, expect disciplined origination, institutional underwriting, and operating leverage in the hold.

Both strategies share the same intelligence infrastructure, the same risk framework, and the same commitment to decisions that are documented, reproducible, and improving with every cycle.

Join the Team

We are looking for people who want to build the next generation of investment firm, not join an existing one.Deal Makers who originate, structure, and execute with proprietary tools instead of slide decks and spreadsheets.Engineers who want their code to move capital, not just ship features.Portfolio Managers who want to work alongside an agent swarm, not replace it.If that sounds like you, get in touch.

Shape the Next Generation of
Intelligent Capital

Board Advisor
Guide strategy, governance, and capital relationships.
Deal Maker
Originate, structure, and execute high-conviction private-market transactions with proprietary tools.
Engineer / AI Builder
Design and build GIC OS. Compress weeks of work into hours through automation and data intelligence.