OpenAI's Denise Dresser: Enterprise customers will be 50% of business by year-end

Key Concepts

• GPT-5.5: The latest OpenAI model iteration, characterized by enhanced reasoning, intuition, and agentic capabilities.
• Agentic Workflow: AI systems that do not just generate text but take autonomous, multi-step actions to complete complex tasks.
• Enterprise Intelligence Layer: The concept of integrating AI across all business functions to unlock data silos.
• Preparedness Framework: OpenAI’s internal safety and security protocols for model development and deployment.
• Codex: An AI-powered coding assistant that serves as a primary use case for agentic software development.

1. Overview of GPT-5.5

Denise Dresser, OpenAI’s Chief Revenue Officer, describes GPT-5.5 as the company’s most intelligent and intuitive model to date. Unlike previous iterations, this model is designed to understand the "shape" of a user's request—interpreting the underlying intent rather than just the literal words.

• Core Strengths: The model excels in long-form reasoning and executing multi-step tasks.
• Practical Application: Dresser illustrates this with a daily workflow example: while a standard model might provide a simple summary of a day’s work, GPT-5.5 understands the user's goal is to finalize tasks, make decisions, and prioritize the following day’s agenda.

2. Enterprise Strategy and Growth

OpenAI is aggressively targeting the enterprise sector, with a goal for enterprise business to account for 50% of its total revenue by the end of the year.

• Codex Success: The company has seen significant traction with its coding assistant, reaching 4 million weekly active users. This serves as a proof-of-concept for "agentic" software development, where AI assists throughout the entire software development lifecycle.
• Cross-Functional Integration: The enterprise vision is to move beyond coding and apply agentic AI to every business function. The goal is to create a unified "intelligence layer" where every employee is assisted by AI, effectively unlocking the value trapped in existing enterprise software ecosystems.

3. Relationship with Existing Software Ecosystems

Addressing concerns regarding the impact of AI on traditional software companies (e.g., Salesforce, Slack, Workday, Databricks, Snowflake), Dresser emphasizes a collaborative rather than disruptive approach.

• Unlocking Data: Dresser argues that AI does not replace these platforms but rather acts as a catalyst to extract more value from them. By using AI to bridge the gap between disparate systems, companies can synthesize collective knowledge across functions.
• Customer-Centric Focus: The narrative presented to enterprise clients is that AI allows them to leverage the data they already possess, turning static systems into active, intelligent workflows.

4. Cybersecurity and Safety Frameworks

When questioned about competition (specifically Anthropic’s "Mythos" model) and cybersecurity concerns, Dresser maintained a focus on OpenAI’s internal standards.

• Preparedness Framework: Dresser noted that OpenAI has operated under a "preparedness framework" for years. This framework is described as "part and parcel" of their development process, focusing on the iterative evolution of models and the democratization of access.
• Defensive Strategy: OpenAI emphasizes supporting "defenders" (security professionals) as a core component of their safety strategy, ensuring that as models become more powerful, the tools to secure them evolve in tandem.

5. Synthesis and Conclusion

The interview highlights a strategic pivot for OpenAI: moving from a consumer-facing research lab to an essential enterprise infrastructure provider. The introduction of GPT-5.5 is the technical engine for this shift, moving the industry from "generative" AI (creating content) to "agentic" AI (executing business processes). By positioning their models as a layer that sits atop existing enterprise software, OpenAI aims to become the connective tissue of the modern digital workplace, prioritizing intuitive reasoning and multi-step task completion as the primary value drivers for corporate clients.