A chaotic pile of old company emails can now be turned into a living "digital twin" that tracks project progress and office culture.
April 24, 2026
Original Paper
Corporate Digital Twins from Email: Using Language Models to Mirror Organizational Life
SSRN · 6632438
The Takeaway
This new pipeline uses language models to extract project milestones and cultural vignettes from raw message archives. It effectively resurrects the operational history of a firm, allowing anyone to query how decisions were actually made. Managers can see a structured map of how teams functioned rather than just guessing from memory. This turns static, forgotten data into a valuable asset for organizational planning and institutional memory. It also raises significant questions about privacy and the permanence of workplace communication. Every internal email now has the potential to become part of a company's living historical record.
From the abstract
<span>We propose corporate digital twins — structured, queryable, and dynamic computational mirrors of organizations — built entirely from internal email archives using large language models (LLMs). Just as digital twins in engineering replicate physical systems for monitoring and simulation, a corporate digital twin replicates the social, operational, and strategic fabric of a firm. We demonstrate this concept on the Enron email corpus (345K emails, 150 employees, 1997–2002), constructing a mul