Pikit (Pikit.ai)

Pikit (Pikit.ai) — Byproduct dataset of in-person homebuyer behavior at model homes and apartment leasing offices: per-lot/per-unit plan views, side-by-side plan rankings, Love/Like/Not-It reactions with free-form notes, dwell time per floor plan, plan-by-plan conversion-to-lead, logged buyer objections (location / lot / home / price) with verbatim quotes, and captured lead records tied to the exact unit reacted to.. Small but high-density: single-founder, pre-scale proptech with a narrow deployed footprint, yet each deployed community produces a dense, uniquely labeled behavioral record (per…

Pikit (Pikit.ai) offers (Alternative, Sentiment, Reference) — Est. ~150-200 toured units and ~600-1,000 discrete plan-interaction events per community per month, each event carrying dwell time, reaction label and note; total corpus currently in the low tens of thousands of labeled interactions across pilot communities (inferred from 148 tours/week snapshot × number of pilot communities).

Consumer-preference modeling for residential product design (which floor plans and finish packages to build in the next phase); price-elasticity and objection-cause modeling for new-home pricing; recommendation/ranking training for housing search and AI answer engines; retail-sales dialogue datasets (buyer objection → agent response pairs) for sales-coaching LLM fine-tuning and RLHF; site-selection and amenity-demand models for multifield developers.

The data is with 2 years of history.

Coverage spans US; Homebuilding, Real Estate Services; alternative, sentiment, reference; equities.