GSK Stockmann

Due diligence is one of the most resource-intensive parts of transactional work at GSK Stockmann, with teams routinely reviewing hundreds of contracts and supporting documents under tight deadlines. More than 60% of the firm's lawyers have frequent touch points with diligence procedures. Manual review consumed significant lawyer time and constrained capacity for higher-value strategic advisory work.

GSK Stockmann partnered with Harvey and hosted a week-long offsite in August 2025 to co-design generative AI workflows spanning the full corporate and real estate due diligence process—from scope determination through report delivery. The solution combined conditional workflows to spot missing or misfiled documents, human-in-the-loop risk validation with Harvey's citation engine, web search against statutory/regulatory and public sources, and workflow playbooks with contextual system prompts (company size, jurisdictions, debt facilities).

Structured M&A, PE, VC, and real estate diligence saw initial time savings of 15–20%, with higher savings on unstructured data rooms. Red-flag summaries that previously took days were generated in hours, and AI-driven spotting of misfiled documents reduced unnecessary back-and-forth with targets. Junior and mid-level lawyers accelerated risk prioritization skills, freeing senior associates for more strategic work and enabling broader analysis beyond the data room.

• • Codifying firm standards with structured prompting and a prompt-creation tool ahead of immersion workshops was pivotal to successful AI process redesign.
• • Match tool complexity to the task: simple shared queries for one-offs; Vault analysis tables and custom workflows with conditionality and human-in-the-loop steps for complex diligence.
• • Immersive, interdisciplinary collaboration (lawyers, legal engineers, developers, and the vendor) shifts teams from ad-hoc experimentation to structured, everyday AI adoption.