Gigi spent seven years at Google before leaving to start a research company.
Pai isn't any old research company; it specializes in the creation of "digital twins".
These are digital replicas based on real consumers.
Pai interviews these consumers and builds digital versions of them. The real respondents get paid every time their digital twin is used, and the real consumers are interviewed regularly so the digital twin can be constantly updated.
Gigi imagines a world where every marketer has their digital twins available to them 24/7, so they can be questioned as part of the overall workflow, vs. a separate and time-consuming exercise.
In the conversation, we cover a ton of ground
The "Say/Do" Gap
The core business problem is not a lack of data. It is the gap between what people say in research and what they actually do in the market. That is framed as the founding problem Pi is trying to solve.
Grounding in Truth
Digital twins only work if they are grounded in real, high-quality data: interviews, category context, and purchase history. Without that, the whole machine turns into a cardboard dragon.
The AI Needs to Apply Moderator Craft
The AI interviewer who interviews the real respondents needs to replicate “the best interviewer we know” by probing, challenging, and reading for inconsistency.
Research Reports Come Alive
Instead of letting segmentations and old research rot in decks, turn them into something teams can query, simulate with, and use in day-to-day decisions. Static deck becomes a living customer base.
The Challenges
Bias, drift, memory limits, and context windows are real challenges that need to be solved.
