RideTalk

An interruptible, long-memory AI voice companion built for commuting

Problem

A user commuting an hour a day by e-bike wants to talk with AI through Bluetooth earbuds: real-time interruption, long-term memory that summarizes into lasting notes, follow-up on casually mentioned to-dos, and a model with its own opinions instead of constant agreement. Mainstream voice assistants feel smooth but sycophantic and forgetful, and none are designed for noisy, hands-free commuting.

Solution

A voice-first mobile app with earbud-button wake, strong barge-in, and a noisy-environment conversation mode. Each session is distilled into long-term memory cards that can be reviewed and searched; mentioned tasks and promises flow into a reminder list; users pick a persona, including a coach that pushes back.

Users

Commuters who spend 30+ minutes daily on a bike, in a car, or on foot and want to think and capture ideas on the go.

MVP

Earbud wake plus real-time interruptible voice chat; auto-generated memory summary cards after each session; to-do extraction with local reminders; 2–3 persona presets; no social or content community features.

Revenue

Free tier plus monthly subscription under $20, tiered by voice minutes or memory storage; optional one-time local-model edition.

Why now

Real-time voice models and earbud ecosystems are mature, and commuting is a high-frequency need that general assistants ignore; a solo developer can win on interruption, memory distillation, and persona details.