Opening Bell Ventures released Shelly, an AI that interviews employees and customers, then tests their answers against company records. A phone demo is open to anyone.
A New York reader might call it an AI auditor that makes phone calls. Opening Bell Ventures, a consulting firm based in Austin, Texas, announced Shelly on October 22, 2026. She talks to employees and customers, then checks what they said against a company's own policies, reports and data. The company says she shows where the two match and where they do not.
Most AI tools record a call, write it down or summarize it. Shelly is meant to go a step further. She asks questions, follows up when an answer is vague or conflicting, and then looks at the records. She can speak by phone or join Microsoft Teams and Zoom meetings.
The company says every finding links back to the conversation, document or data point behind it. She also keeps going after the report, checking in with the people responsible for fixing problems and tracking progress.
The pitch rests on a simple idea. Policies describe one process, while the team doing the work may follow another. A dashboard may say an issue is closed when the person who owns it knows it is not. CEO Avi Kalderon put it this way: "Every organization has two versions of the truth: what people say happens, and what the records show."
Shelly is invitation-only for now. Opening Bell Ventures is aiming at mid-size and large organizations where rules and complexity make proof important. The company names these fields:
Listed uses include getting ready for audits, tracking fixes after problems are found, and talking with customers about their experience. She can also answer employee questions about company policies and send requests to a person when needed.
The company says Shelly works within set access permissions and confirms a person's identity before sharing restricted information. She asks for confirmation before making any authorized change. Client information is logically isolated, and each interaction and action is logged with a timestamp. Anything outside her assigned job goes to a designated person.
The announcement does not give a price, the number of organizations in the beta, or any results on how accurate her findings are. It also does not explain what happens to the phone number a visitor enters for the demo. Anyone curious can try the demo at shelly.openingbellventures.com, where Shelly calls the number entered and talks through a business problem.
Shelly is an AI that compares what people say with what records show, but pricing and accuracy are unknown and access is invitation-only for now.
Visa tested 22 AI models on realistic shopping tasks. They passed just over half, and often bought the wrong thing anyway.
Read more: Visa tested 22 AI shoppers, and most still make costly mistakesDARPA wants AI to hunt for design flaws in stock, bond, energy and cloud markets, so enemies cannot use them to cause economic damage that reaches households.
Read more: Pentagon wants AI to find weak spots in stock and energy marketsTuring Award winner Richard Sutton says more computing power alone will not make AI truly smart. Big AI companies disagree, which matters for anyone betting on AI.
Read more: AI pioneer says piling on computing power won't make machines smart
