AI Forms & Analytics

A feedback platform that watches how people fill in a form, scores their attention and checks free-text answers with a local LLM.

The project is ranked in the Top 100 of the Moscow Innovation Cluster.

Behavioral Telemetry

The platform captures rich behavioral signals as users interact with forms: mouse movements, answer selection patterns, field focus duration, and scroll behavior. This telemetry feeds into the attention scoring model.

Attention Scoring

A hybrid CNN+MLP architecture processes behavioral telemetry to generate an attention score for each response, distinguishing engaged respondents from those rushing through forms.

AI-Powered Validation

Beyond behavioral telemetry, open-ended responses are validated using a locally-hosted custom AI model, ensuring data quality without sending sensitive information to external servers. The LLM detects nonsensical answers, spam, and low-effort responses in real time.

Key Highlights

• Custom LLM: filters out low-quality open-ended responses
• CNN+MLP scoring: AI attention model from behavioral signals
• Real-time telemetry: mouse, scroll, focus, and hesitation tracking
• Privacy-first: all AI processing runs locally