Unlike previous technologies, ChatGPT’s launch created immediate, widespread pressure on biopharma executives. Prompted by their boards and even families, they recognized the potential to leapfrog years of development, rapidly elevating AI on the corporate agenda despite concerns about data privacy and IP.

The opportunity cost of building custom internal AI can be massive. By the time a multi-million dollar project is complete, off-the-shelf tools like ChatGPT are often far more capable, dynamic, and cost-effective, rendering the custom solution outdated on arrival.

The conversation around AI in healthcare often focuses on patient-facing chatbots. However, the more significant, unspoken trend is adoption by clinicians themselves. As of last year, two out of three American doctors were already using AI for administrative tasks, translation, and even as a 'wingman' for clinical diagnosis.

The widespread use of AI for health queries is set to change doctor visits. Patients will increasingly arrive with AI-generated analyses of their lab results and symptoms, turning appointments into a three-way consultation between the patient, the doctor, and the AI's findings, potentially improving diagnostic efficiency.

OpenAI's launch of ChatGPT Health, which integrates medical records, signals a clear strategy to move beyond general-purpose APIs. Foundation model companies are now building specialized, vertical-specific products, posing a direct threat to "wrapper" startups that rely on the underlying models' existing capabilities.

The relationship between AI startups and pharma is evolving rapidly. Previously, pharma engaged AI firms on a project-by-project, consulting-style basis. Now, as AI models for drug discovery become more robust, pharma giants are seeking to license them as enterprise-wide software suites for internal deployment, signaling a major inflection point in AI integration.

For companies given a broad "AI mandate," the most tactical and immediate starting point is to create a private, internalized version of a large language model like ChatGPT. This provides a quick win by enabling employees to leverage generative AI for productivity without exposing sensitive intellectual property or code to public models.

Mollick warns against the common first AI project: a Retrieval-Augmented Generation (RAG) chatbot for internal documents. These custom projects are expensive, and their functionality is often quickly surpassed by cheaper, more powerful off-the-shelf models, resulting in a poor return on investment.

The creation of ChatGPT Health was not a proactive pivot but a direct response to massive, organic user behavior. OpenAI discovered that 1 in 4 weekly active users—over 200 million people globally—were already using the general purpose tool for health queries, validating the immense market demand before a single line of dedicated code was written.