Generative AI Reshapes B2B Buying Decisions

A Harvard Business Review feature by IMD's Amit Joshi and Ivy Buche and GSK's Caroline Schwaer (published June 12, 2026) argues that generative AI has moved B2B discovery, evaluation, and recommendation into AI-mediated environments that vendors neither own nor fully control - a structural shift, not a marketing problem. The article illustrates the effect with an oncology scenario: a newly approved cancer drug with strong Phase 3 trial results can be recommended less prominently by an AI assistant than an earlier, more-discussed competitor with more public real-world evidence, even when the new drug's clinical data is stronger.

HBR reports that the clinical decision-support assistant OpenEvidence is used daily by more than 40% of U.S. physicians, with usage up roughly sevenfold from about 2.6 million sessions in December 2024 to a much larger January 2026 total, per the company's internal figures cited in the article. Fierce Healthcare's independent January 2026 reporting corroborates that OpenEvidence is used daily by more than 40% of U.S. physicians across 10,000-plus hospitals and medical centers, lending outside support to the scale HBR describes. The article also notes that OpenAI and Anthropic increased healthcare focus in early 2026, according to HBR.

The practical takeaway for teams building search, retrieval, or content pipelines: AI-mediated discovery depends on retrieval quality, index coverage, and machine-readable, evidence-rich content, so structured data, metadata, and RAG-friendly formatting become go-to-market engineering priorities, not just documentation nice-to-haves. This is not healthcare-specific - any domain where an assistant synthesizes a recommendation will privilege sources that are discoverable and structured for machine reading, an industry-wide pattern rather than a claim about any single firm's roadmap.

Adoption and query-volume trends for vertical assistants and their measured effect on procurement decisions; whether organizations publish more structured, machine-readable evidence to influence algorithmic retrieval; and regulatory or auditability scrutiny of provenance and explainability in regulated sectors like healthcare.