OpenAI on OpenAI: Stacie Faggioli, Business Finance Officer Applications, OpenAI

Key Concepts

• AI-Native Finance: Designing workflows around AI agents rather than bolting them onto existing processes.
• Headcount Leverage: Achieving higher output with a significantly smaller team compared to industry peers.
• Code Interpreter (Codeex): A tool enabling non-technical staff to perform data analysis, automation, and front-end interface creation.
• Agentic Workflow: Using autonomous AI agents to orchestrate tasks across multiple systems (procurement, credit risk, contract review).
• Democratized Innovation: Empowering employees closest to the data to build their own AI solutions via hackathons.

---

1. Core Principles of the OpenAI Finance Team

Stacy Fajoli outlines three pillars that guide the finance organization’s evolution:

• AI-Native by Design: Moving beyond simple automation to fundamentally reimagining workflows. This includes rethinking organizational charts to center around AI agents.
• Headcount Leverage: A PWC assessment revealed that OpenAI’s finance team is only 20% the size of comparable technology peers, proving that the right tooling allows for "doing more with less."
• Deploy Early and Iterate Fast: Prioritizing speed over waiting for "final" versions of technology, allowing the team to grow organically alongside the AI.

2. Organizational Structure

The finance team is divided into three pillars:

1. Strategic Finance: Capital allocation, fundraising, and growth planning.
2. Finance Operations: The "engine" of the team (accounting, revenue collection, tax, and monthly book closing).
3. Enterprise Fintech: Manages data platforms and systems.
   • Strategic Note: OpenAI embeds engineers directly within the Fintech pillar to work side-by-side with finance subject matter experts, eliminating the bottleneck of waiting for IT departments to fulfill requests.

3. Individual Productivity Tools

Investor Relations Agent

• Application: Used during two major equity raises ($40B and $122B).
• Methodology: Trained on internal data and the tone of a public-company IR professional.
• Impact: Provided data-grounded, consistent answers to investor diligence requests in minutes, saving hundreds of millions in advisory fees. It also scaled the "equity story" across recruiting and executive teams.

ChatGPT for Excel

• Application: Automating complex financial modeling (e.g., LBO models).
• Process: The tool structures the analysis, writes formulas directly into the workbook, and allows for traceable, auditable assumptions.
• Efficiency: Reduced tasks that previously took hours or days to approximately 10 minutes.

Codeex (Code Interpreter) for Data Analysis

• Marketing ROI: Uploaded raw marketing spend data to generate an ROI dashboard that identifies channel saturation and suggests budget reallocations weekly.
• Sales Insights: Analyzed Gong transcripts and customer emails to track if sales reps were pitching new products, providing leading indicators before the quarter ended.
• Executive Reporting: Automated the creation of complex "compute margin" slides by reconciling infrastructure telemetry with accounting rules, saving days of manual work per month.

4. Organizational Agentic Workflows

The team has deployed specialized agents to handle routine, high-volume tasks:

• Procurement Agent: Deflects ~60% of travel and procurement policy questions.
• Credit Risk Agent: Replaced manual research with automated scoring embedded directly into the CRM.
• Contract Review Agent: Ingests bulk agreements to flag non-standard terms, ensuring compliance with GAAP/ASC 606 without needing to scale the accounting team linearly.
• Vendor Risk Agent: Automates risk reporting and embeds scores into procurement software for immediate approval/escalation.

5. Key Arguments and Perspectives

• The "AI Mindset": Fajoli argues that success is not just about the tools, but about conditioning the team to ask, "How can I use AI to make this task easier?" before starting any project.
• Democratization: Fajoli emphasizes that the most impactful innovations did not come from leadership, but from employees "plumbing the systems every day." By providing access to tools like Codeex, the team self-innovated.
• Human-in-the-Loop: Despite high levels of automation, the team maintains rigorous QA and "evals" (evaluations) to ensure data accuracy and pass "sniff tests" before presenting to the CFO or board.

6. Synthesis

The OpenAI finance team serves as a blueprint for the "finance team of the future." By embedding engineers within finance, prioritizing AI-native workflows, and democratizing access to powerful tools like Codeex, they have achieved extreme operational efficiency. The primary takeaway is that AI should not be treated as an add-on, but as a foundational element that allows a small, agile team to handle the financial complexities of a rapidly scaling organization.