Demonstration • Confidential

Pineal Capital
AI Operations Engine

A purpose-built AI system that works alongside your team — seeking positive absolute returns through deep-dive research into deep value, catalyst-driven and secular growth investment opportunities.

7 AI Skills
3 Autonomous Agents
25+ Python Modules
75 Files Built
What is this?

Your fund's brain — powered by AI

We've built a custom AI system specifically for Pineal Capital Fund I. Think of it as a highly trained analyst that never sleeps, never forgets, and always follows your investment principles.

It doesn't replace your team — it makes your team faster. Every output follows your 5 Investment Principles and the 3 U's framework automatically.

  • Builds investment theses from scratch in minutes, not days
  • Monitors every catalyst across your portfolio continuously
  • Generates IC-ready memos with full compliance checks
  • Challenges every thesis with a built-in devil's advocate
  • Enforces your risk limits (40/60) automatically
  • All outputs clearly tagged — nothing presented as final without review
PINEAL AI ENGINE THESIS BUILDER CATALYST TRACKER RISK MONITOR IC PAPER VALUATION SCREENER COMPLIANCE CHECKER
Foundation

Built on Your 5 Investment Principles

Every AI output is governed by these principles. They're not guidelines — they're hard-coded rules the system cannot override.

1

Capital Preservation

Flags binary outcomes. Rejects secular decline. First cut is cheapest.

2

Margin of Safety

Only proceeds when odds favour us. Quantifies the unknowns discount.

3

Asymmetric Risk/Reward

Seeks 3x+ setups. Rejects symmetric risk. Shows the payout math.

4

Understand the Business

No shortcuts. Cash flows and market book values. Predictable economics.

5

Efficient Portfolio Opt

Uncorrelated returns. Tactical management. Reduce cost basis systematically.

Capabilities

7 Specialist AI Skills

Each skill is a purpose-built workflow. Ask the AI to do something and it automatically selects the right skill, follows the right process, and produces the right output format.

🔎

Thesis Builder

Constructs full investment theses from scratch: applies the 3 U's filter (Under Researched, Under Owned, Under Performed), analyses the business, maps catalysts, builds equity bridges, and calculates margin of safety — all in one pass.

Core Research

Catalyst Tracker

Maps every catalyst for every position: M&A, management changes, capital returns, strategic pivots, earnings inflections, technical breakouts. Tracks timelines and flags value trap risk when catalysts stall.

Monitoring
📈

Valuation Screener

Runs Sum-of-the-Parts (SOTP) analysis, relative value vs peers, and equity bridge calculations. Applies the "unknowns discount" methodology to determine entry prices with built-in margin of safety.

Valuation
🛡

Risk Monitor

Enforces your risk limits automatically: 40% single stock, 60% sector maximum. Tracks drawdowns against stop-losses, monitors hedge ratios, and alerts when positions need review. Three-factor position sizing: liquidity × volatility × payout.

Risk Management
📄

IC Paper Generator

Produces Investment Committee memorandums that consolidate thesis, valuation, catalysts, risk, and compliance into a single structured document. Follows the fund's pre-investment → IC → live management process exactly.

Documentation

Compliance Checker

Three-track compliance: QIAIF fund structure (CBI/AIFMD), investment restrictions (position limits), and ESG exclusion screening (Ottawa Treaty, coal, weapons, authoritarian regimes). Blocks non-compliant investments automatically.

Regulatory
🚀

Session Bootstrap

Initialises every session: loads fund context, reads prior learnings, classifies the task at hand (Research, Trading, Risk, or Compliance), and routes to the correct specialist skill. Ensures continuity between sessions.

Operations
Always On

3 Autonomous Agents

Unlike skills (which you ask for), agents work independently in the background — monitoring, alerting, and challenging without being prompted.

☀️

Morning Briefing

Daily at 07:00

Overnight market moves, portfolio position updates, catalyst calendar for the week, macro regime assessment, and priority actions for the day. Your daily intelligence summary.

🚨

Catalyst Sentinel

Continuous

Watches every active position for catalyst events: earnings releases, M&A announcements, board changes, regulatory decisions, insider transactions, technical breakouts. Alerts instantly.

🧠

Thesis Challenger

Weekly + On-Demand

Plays devil's advocate against every live thesis. Generates bear-case memos, challenges assumptions, identifies where margin of safety has eroded. Verdicts: Holds / Weakening / Broken.

Process

How the System Works

From first screen to live portfolio management — the AI follows your investment process at every stage.

1

Screen & Filter

The AI scans a universe of European mid-caps through the 3 U's filter: Under Researched (<5 analysts), Under Owned (<30% institutional), Under Performed (>30% below highs). Only stocks passing at least two criteria advance.

2

Deep-Dive Research

The Thesis Builder skill analyses the business model, financials, management quality, and competitive position. Simultaneously, the Valuation Screener runs SOTP analysis and the Catalyst Tracker identifies value-unlocking events.

3

Challenge & Validate

Before anything reaches IC, the Thesis Challenger agent generates a rigorous bear case. The Risk Monitor checks portfolio fit (40/60 limits), and the Compliance Checker runs ESG exclusion screening.

4

IC Paper Generation

All research, valuation, catalysts, risk assessment, and compliance output is consolidated into a structured Investment Committee Memorandum — complete with your 8-item investment checklist and recommendation.

5

Live Portfolio Management

Post-investment, the Morning Briefing agent provides daily intelligence, the Catalyst Sentinel monitors events in real time, and the system continuously reassesses every thesis against evolving conditions.

Worked Example

UniCredit Case Study

Here's how the AI would have processed one of Pineal Capital's historical investments — demonstrating every stage of the system.

UniCredit S.p.A. (UCG.MI)

Pan-European commercial bank • Market Cap >€60bn • Net Income €9.5bn

0.3x P/TNAV at Entry
57% Discount to Peers
+70% Share Price Rally
+30% Outperformance vs Peer

3 U's Assessment: All PASS

Under Researched: underappreciated turnaround narrative. Under Owned: low conviction after 3yr underperformance. Under Performed: cheapest large-cap European bank at 0.3x P/TNAV.

Equity Bridge (how the AI maps value creation)

0.3x P/TNAV Entry Price
New CEO Orcel Catalyst
ECB Rates Profitability Shift
Buybacks Capital Return
+70% Outcome
Technology

Open-Source Toolkit

The system is built on the most trusted, widely-used tools in quantitative finance — used by institutions worldwide. All open-source, all battle-tested.

Category Tool GitHub Stars What It Does
Data Science pandas + numpy + scikit-learn 41K+ Data analysis, financial modelling, factor analysis
Portfolio PyPortfolioOpt + CVXPY 9.8K Portfolio optimisation, risk parity, Black-Litterman
Technical Analysis pandas-ta 5K+ 130+ technical indicators (RSI, MACD, Bollinger, etc.)
Market Data yfinance 14K+ Real-time and historical market data
Analytics pyfolio + plotly 17K+ Performance tear sheets, interactive dashboards
AI Engine Claude (Anthropic) Multi-agent orchestration, thesis generation, IC papers
Repository Structure

What's in the System

The entire system is organised into clear modules. Here's a plain-language map of what each folder contains and why it matters.

💡

CLAUDE.md — The Fund Brain

A 148-line instruction file that loads every time the AI starts. Contains your team details, 5 Principles, 3 U's framework, risk limits, and hard rules. This is what makes the AI think like Pineal Capital.

📚

research/ — Analysis Engine

Four sub-modules: fundamental (financial models, SOTP, peer comparison), macro (regime detection, sector screening), technical (chart analysis, entry/exit), and catalyst (mapping, event monitoring).

📊

portfolio/ — Risk Management

Position sizing (liquidity × vol × payout), risk parameter enforcement (40/60 limits), drawdown management with stop-losses, hedge manager (tail hedges, FX, derivatives), and exposure tracking.

🤖

agents/ — Multi-Agent System

Five specialist AI agents that can work in parallel: Thesis Analyst, Risk Guardian, Macro Analyst, Catalyst Hunter, and Devil's Advocate. Coordinated by a central orchestrator.

📋

ic_papers/ — IC Templates

Ready-to-use templates for investment theses and IC memos. Includes a worked UniCredit case study showing exactly what the system produces for a real investment situation.

📑

notebooks/ — Interactive Demos

Four Jupyter notebooks you can run: SOTP valuation, catalyst mapping, risk dashboard, and multi-agent orchestration. Visual, interactive demonstrations of each capability.

Guardrails

Built-in Safety & Compliance

The system has hard-coded rules that cannot be overridden. These protect the fund, the team, and your investors.

Anti-Hallucination

Every data point is tagged: [VERIFIED], [UNVERIFIED], [ESTIMATED], or [PENDING RESEARCH]. The AI will never fabricate returns, Sharpe ratios, or financial figures.

ESG Exclusions

Ottawa Treaty (mines/munitions), coal ≥10%, weapons >10%, authoritarian regimes, controversial weapons, child/forced labour. Absolute exclusions cannot be bypassed.

Irish Regulatory

All outputs respect CBI/QIAIF/AIFMD rules. SFDR Article 6 compliance. The system flags any output approaching regulated financial advice territory.

Human Review

Every output is tagged [AI-DRAFT] until a named partner signs off. The AI assists — it never makes investment decisions autonomously.