The Business of Intelligence

Key Concepts

• Code Red: A company-wide initiative to force intense focus on a specific area or problem.
• Enterprise AI: The application of artificial intelligence within large organizations to improve productivity, efficiency, and innovation.
• ChatGPT: A user-level AI product designed to unlock productivity for teams.
• APIs (Application Programming Interfaces): Interfaces that allow developers to build with AI models in a raw, direct access manner.
• Codeex: A user-directed AI software engineer tool that leverages an organization's codebase.
• Agentic Systems: AI systems capable of performing complex, long-horizon tasks autonomously.
• Thrive Holdings: A company that partners with OpenAI to explore organizational transformation through AI, combining VC, company building, and product development.
• Replatforming: A fundamental rethinking and restructuring of business processes and systems to integrate AI effectively.
• Intent CX: A T-Mobile initiative to reimagine their contact center as an information hub for understanding customer intent and driving engagement.
• Moat: A sustainable competitive advantage for a company.
• People Enablement: The process of training and supporting employees to effectively use new technologies like AI.
• Deflationary Impact of AI: The potential for AI to significantly reduce the cost of producing goods and services, thereby increasing organizational output.

Focus on Enterprise AI and Organizational Transformation

The discussion centers on the strategic imperative for companies to embrace AI, particularly within the enterprise. The concept of a "Code Red" is introduced as a method to drive this focus, highlighting its refreshing effect on organizations attempting to manage numerous initiatives at high speed. The speaker emphasizes that this refocus is not just about adopting new tools but about fundamentally rethinking how businesses operate.

Enterprise Product Levels and the "Middle Ground"

OpenAI currently offers two primary levels of enterprise product:

1. User-Level Product (ChatGPT): This product is designed to unlock productivity for teams. A recent enterprise AI report indicated significant benefits, with users reporting saving an hour per day. Key findings from the report include:

   • Message volume increase of 30% per user year-over-year.
   • Total messages up by 8x.
   • A notable 36% year-over-year increase in the use of coding models by individuals in non-technical functions. This level aims to "supercharge" every team member, setting an expectation that individuals can and will achieve more.

2. API-Level Access: This provides developers with raw, direct access to build with AI models.

The speaker identifies a gap in the "middle ground," where powerful, agentic tools are needed to solve long-horizon tasks. These tools should be directed by individuals who understand the work but leverage the enterprise's existing infrastructure, including codebase, data, SOPs, access controls, security protocols, and permissions. Tools like Codeex, described as an "AI software engineer," exemplify this emerging trend.

The "Code Red" and Future Model Improvements

The "Code Red" initiative is intended to accelerate improvements in specific focus areas within AI models over the next few cycles. The speaker anticipates the release of new features and capabilities that will shift the perception of what is possible in both enterprise and consumer applications. The primary focus remains on enterprise AI penetration, moving beyond user-led and developer-led adoption to address the core processes and work that define large enterprises.

Partnership with Thrive Holdings

A significant announcement is the partnership with Thrive Holdings. Thrive Holdings is described as a unique entity that functions as both a venture capital firm and a company builder. They are also an investor in OpenAI.

• Kinship Beyond Investment: The relationship with Thrive transcends a typical investor dynamic. Thrive is deeply interested in OpenAI's technology and its practical application.
• Transformation Through Rethinking: Thrive's approach involves not just integrating AI but fundamentally rethinking the entire organizational structure, combining AI models, organizational change management, and product building.
• Exploring Transformation: The partnership aims to explore how these three elements can be intentionally married to drive genuine transformation. OpenAI contributes energy and excitement about AI capabilities, while Thrive offers expertise in company building and application.
• Addressing "Circular" Deal Perception: The speaker refutes the notion that the deal is "circular," framing it as being in the business of building companies, with AI serving organizations. They distinguish this from a simple SaaS subscription model.

The Nature of Enterprise AI Adoption: Replatforming, Not Just Subscription

The speaker argues that AI in the enterprise necessitates a "replatforming" rather than a simple adoption of a new software feature. This involves a top-to-bottom rethinking of entire business units and their operations. The analogy used with CEOs is to imagine a startup attacking a specific area of their business; this mindset is crucial for understanding the transformative potential of AI. The Thrive partnership is informed by this company-building perspective.

Organic Enterprise Adoption and Categorization of Deployments

Historically, OpenAI's engagement with the enterprise has been largely organic, with customers approaching them rather than OpenAI pursuing specific clients. This was partly because the initial API was built for developers, and ChatGPT was conceived as a research experiment, not an enterprise product.

Deployments are generally categorized into three areas:

1. Team-Level Enablement: Empowering individual teams and users.
2. Product and Customer-Facing Enablement: Enhancing customer interactions and product offerings.
3. Process Enablement: Optimizing internal business processes.

High Acceleration Areas

The highest acceleration is observed in team-level and user-level progress. This is evidenced by:

• Custom GPTs and Projects: These features within ChatGPT now account for a significant portion (20%) of enterprise message volume, representing a 20x year-over-year increase. This surprised the speaker, highlighting a strong desire among users to take control of their workflows and accelerate their own productivity.
• Advanced Capabilities: Users can now leverage AI systems to rebuild entire websites, create high-fidelity visual and media assets (e.g., with models like Sora), and perform long-horizon, end-to-end software engineering tasks (e.g., with coding models like Codeex).
• Agentic Systems: The trend is towards configuring agentic systems to perform these complex tasks.

The speaker considers it a "miss" if teams are not utilizing the best AI models for core productivity, software engineering, media creation, and other functions.

Customer Experience and Process Enablement Examples

• T-Mobile Case Study: T-Mobile is presented as a prime example of rethinking customer experience. They approached OpenAI not just for customer support but to reimagine their contact center as a communication channel for product understanding, benefit delivery, and feedback collection. This led to the development of "Intent CX," an initiative focused on distilling customer intent through AI's ability to get to the core of what customers want in an approachable and warm manner. This demonstrates a process of rethinking the entire customer engagement strategy around AI, not just implementing a technology.
• Complex Business Processes: Areas like supply chain and revenue cycles are also targets for AI enablement.

"Sticky" Areas and Challenges

The most challenging areas for AI implementation are those with a high degree of defined rules and determinism, characterized by:

• Long, End-to-End Processes: Complex workflows that span significant durations.
• Institutional Knowledge: Processes heavily reliant on the deep, often tacit, knowledge of long-term employees (e.g., 20+ years of experience).
• Proficiency Gap: The difficulty of AI climbing the "curve" to match the proficiency of individuals with extensive experience in a specific, nuanced process.

Despite these challenges, these "sticky" areas represent significant opportunities. OpenAI plans to invest in standardizing components for AI implementation in these domains, anticipating future models that can learn these processes rapidly.

OpenAI's Moat and Growth Strategy

OpenAI's valuation of $500 billion is acknowledged, but the focus remains on creating enduring value for customers as their true "moat." Drawing from experience at Y Combinator, the speaker emphasizes that startups win by identifying gaps where large, incumbent companies have high customer dissatisfaction and then relentlessly focusing on customer value.

For OpenAI, this means proving that AI solutions are superior to existing methods in solving problems and creating value for businesses and their customers. The excitement around AI is justified, but continuous validation through customer success is paramount.

Applying AI in the Enterprise: A 90-Day Plan for CEOs

For a CEO starting over with AI in a Fortune 500 company, the emphasis is on a people-led approach.

• AI as Leverage: AI is viewed not as a job replacement tool but as a way to augment existing employees, providing them with significant leverage.
• Organizational Rethink: Implementing AI is akin to adding a substantial number of new people to an organization, requiring changes in processes, systems, and how work is approached.
• Building Familiarity: A process of building familiarity with AI tools and capabilities is essential and takes time.
• People Enablement: Significant investment is made in enabling people to use AI effectively, alongside the AI enablement itself.

The Deflationary Impact and Organizational Capacity

AI's deflationary impact means organizations can do significantly more. The cost of producing software, media, products, or customer interactions can be reduced by factors of 10, 100, or even a million. Consequently, the number of things an organization can achieve in theory should increase proportionally. This requires mastery and accumulation of wins over time.

Personal Perspective: Parenthood and the Future of Human-Computer Interaction

The speaker shares how the recent birth of his child has provided a new perspective. The personal time dedicated to his daughter offers a meditative balance to his intense focus on OpenAI.

• Generational Shift in Computing: Children growing up today will have a fundamentally different relationship with computers. They may not conceive of traditional graphical user interfaces (GUIs) and will intuitively interact with systems through natural language.
• Conversational AI as the Norm: The expectation for children will be that computers are conversational partners, capable of understanding and responding to spoken requests. This mirrors the natural development of verbal communication in children.
• Meeting Humanity Where It Is: This shift represents a long-held technological aspiration: creating systems that meet humans at their current level of understanding and interaction, moving beyond complex abstractions. The speaker is excited about this future where technology is more intuitive and accessible.