AI twins start answering emails and joining meetings for executives

The newest status symbol in the executive suite is not a bigger office but a second self. AI twins are beginning to answer messages, sit in meetings and carry the voice of top leaders into rooms where those leaders are no longer physically present.

Zoom chief executive Eric Yuan laid out that future in June 2024, describing digital twins that could attend meetings on a user’s behalf and eventually make decisions for them. He also sketched a workplace where those AI versions would handle email, pushing the idea beyond simple scheduling help and toward a more automated model of management. What once sounded like a far-off product demo is now being built into everyday tools.

By 2025, the concept had already moved into live use cases. Professors’ AI twins were helping students with questions outside normal hours, extending access after classes and office hours ended. In the corporate world, Klarna chief executive Sebastian Siemiatkowski was cited as having sent an AI double to speak during a financial update, while Otter.ai chief executive Sam Liang said a meeting-attending avatar prototype could be ready later that year. Liang has also described a “Sam-bot” meant to eventually take his place in meetings.

The shift accelerated in February 2026, when Seattle startup Read AI launched a product called Digital Twin, branded Ada. The tool works through email to help schedule meetings, answer questions and keep conversations moving. That matters because it moves AI presence from a speculative pitch into an actual workplace product, one that can sit between employees and the people who manage them.

The appeal is obvious to executives who want to compress their calendars and appear available without being fully reachable. But the tradeoff is more than convenience. Once leaders are represented by AI versions of themselves, workers may not know whether an answer reflects the person in charge, a delegated system or a machine trained to sound like both. Questions about consent, authority and the right to shut down an AI twin are becoming central to how these products are judged.

There is also a security cost to normalizing boss-clones. If employees get used to receiving instructions, approvals and conversational responses from AI versions of colleagues and managers, they may become more vulnerable to impersonation and scams. The workplace may look faster and more efficient, but the distance between power and accountability could widen just as the interface becomes friendlier.