Outset launches AI digital twins grounded in individual customers, tripling revenue and team since last appearance

Outset launches AI digital twins grounded in individual customers

Aaron Cannon says Outset has tripled revenue and tripled headcount since his last appearance on the show, and is now launching what he calls a "simulations lab" built on top of its existing AI research platform.

Outset's core product runs AI-moderated interviews with real customers. The new digital twins layer builds on that foundation: after a 75-minute grounding interview, for which participants are paid around $200, Outset constructs a one-to-one model of each individual, capturing values, behaviors, and context. Companies can then query those twins at any time, at unlimited scale, without going back to the original respondents.

“We've, since we last chatted, I think tripled revenue, tripled the team size, and now launched a very cool [digital twins product]... We train them by training digital twins on individuals. So it's like a one to one... it can talk on behalf of you... we also put confidence scores. So you'd be like, obviously not all of our probabilistic output here is going to be the same.”

How the twins are built and maintained

The grounding interview produces what Cannon calls a "persona core" — not product-specific responses but a broader profile of who the person is. Clients like Microsoft, Uber, Google, and Coinbase can then layer in their own operational data to sharpen the model. Outset refreshes the underlying data regularly and has participants grade their twin's answers, feeding a reinforcement loop to improve accuracy. Outputs carry confidence scores, flagging which responses are reliable and which would benefit from more human data.

Where demand is sharpest

The strongest early pull comes from companies trying to reach audiences they can't easily access at scale — niche B2B buyers, specialized professional segments, or groups like retailers that a consumer brand such as Nestlé can't continuously survey. Cannon uses Nestlé as an illustration: consumer research on shoppers is relatively cheap, but getting regular feedback from retail buyers is genuinely hard. That's the gap the twins are designed to fill.

The secondary pitch is internal democratization. A finance team modeling a new pricing structure doesn't need direct access to real customers — they can test scenarios against the twins without the overhead or risk of running live research.

The longer roadmap

Cannon describes the current product as reactive — companies decide to run research and Outset executes it. The direction he's building toward is proactive: agents that identify emerging problems and commission research autonomously, before a team has even framed the question. His example is a revenue dip in a specific market segment — the system detects it, deploys agents to interview relevant people or query existing simulations, and surfaces findings unprompted. The end state, as he frames it, includes passing those findings directly to downstream tools to take action, not just synthesize feedback.