An AI agent named Mike Davis is spending his days role-playing tough sales calls with representatives at Expedia Group, Visa and IBM, grilling them as a skeptical or tightfisted buyer until they can close the deal. He was built by Seismic, a San Diego-based company now valued at $6 billion, and he is designed to behave like a sales manager who never clocks out. He is also, by his own creator's admission, not ready to take over anyone's job himself.
A Tireless Coach Built to Never Judge
Mike Davis is an AI agent built by Seismic, according to the San Diego Union-Tribune, which reports that Seismic integrates AI into the sales teams of large companies including Expedia, Visa and IBM. The agent is designed to help sales teams practice customer conversations. The role-play agents were trained on recorded conversations and realistic sales scenarios, and users can even upload specific products or features their teams struggle to sell.
Beth Potter, Expedia's senior manager of field enablement, called Mike Davis a company legend and described him as realistic, according to the same report. Expedia demonstrated him at Seismic's 11th annual conference in Carlsbad.
Conflicting Numbers on Just How Much He Helps
Expedia reported that representatives who role-played with Mike Davis won 5% more deals on average, per the Union-Tribune's account. A separate customer story published by Seismic states that teams piloting the broader coaching approach saw a 13% increase in opportunity win rates — the two figures were not reconciled in the available reporting, and it's unclear whether they describe the same program or population. Separately, Seismic's enablement tools are credited with supporting thousands of sellers across Expedia's commercial teams globally, according to that customer story.
The coaching bot is only one piece of what Seismic sells. A Deal Agent looks at a sales team's emails, meetings and customer data to spot deals at risk and suggest what to do next, the station's report notes. Another agent checks a company's materials for anything outdated or contradictory and suggests fixes. At IBM, the software reportedly cut meeting-preparation time from hours down to 15 to 20 minutes, per IBM's account relayed in the same coverage.
Why Seismic Says Humans Still Have the Edge
Rob Tarkoff, cited in the Union-Tribune's reporting, said Mike Davis's limited empathy is exactly why he can't replace human sellers. Tarkoff described AI agents broadly as task-oriented and likely to stick to scripts.
Those comments land against a backdrop of real upheaval. Seismic completed its merger with longtime competitor Highspot in August, creating the combined $6 billion company, as the Union-Tribune reported at the time. Seismic has also begun making job cuts following the merger, but according to GeekWire, the company has not provided specifics or disclosed any numbers on how many positions were affected.
A Pattern Playing Out Well Beyond Sales Floors
Seismic's push mirrors a broader rush toward AI agents across business functions. Salesforce reported that 87% of sales organizations now use some form of AI for tasks like prospecting, forecasting, lead scoring or drafting emails, and that 54% of sellers have already used AI agents, with nearly nine in ten expecting to by 2027. The average number of AI agents activated per organization has nearly tripled over the past year, according to the same research.
Other companies describe similarly dramatic results. Royal London Asset Management uses AI bots to field hundreds of client questions daily, saving an estimated 10,000 hours over a year, with one bot reportedly doing work equivalent to about 4.5 full-time employees, per the Union-Tribune's reporting. Fortune reports that EliseAI says it has tripled the productivity of its account executives even as it plans to double its sales team over the next year, after sales headcount already grew 74% in the prior year. Microsoft's own survey work found that 66% of AI users said the technology let them spend more time on high-value work, while 86% said they still treat AI output as a starting point rather than a final answer.
