Privacy by design: Building Earmark

2026-02-09 · Source: AssemblyAI · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Cybersecurity & Data Privacy, Software Development & Engineering · Depth: Intermediate, quick

Summary

Earmark, a voice AI product, emphasizes privacy by design, offering a "temporary mode" across all plans that bypasses database storage for transcripts and user data entirely. This feature reflects a strong commitment to handling sensitive voice data by avoiding retention, encryption, and careful consideration of storage duration. Beyond privacy, Earmark's development highlights the importance of a forgiving user experience (UX) for voice AI products. The design philosophy prioritizes minimal interaction, aiming for one-click or automatic operation to initiate conversation capture, and self-resolution for any operational glitches, acknowledging that users often engage with voice AI while focused on other primary tasks like meetings or phone calls.

Key takeaway

For product managers developing voice AI solutions, prioritize privacy from the outset by offering clear data retention controls, such as a no-storage option. Simultaneously, simplify the user experience to a single interaction point for core functions, ensuring the product integrates seamlessly into existing workflows without demanding significant user attention. This approach enhances adoption and trust by respecting user privacy and minimizing cognitive load.

Key insights

Voice AI products require privacy by design and a forgiving, low-friction user experience.

Principles

• Voice data is inherently sensitive.
• UX should minimize user decisions.

Method

Implement a "temporary mode" to prevent data storage. Design for one-click activation and self-resolving errors to reduce user burden during primary tasks.

In practice

• Offer opt-out for data retention.
• Design for single-action initiation.
• Build self-correction into workflows.

Topics

• Voice AI
• Privacy by Design
• User Experience
• Data Retention
• Product Design

Best for: Product Manager, CTO, VP of Engineering/Data, Entrepreneur, AI Product Manager, AI Engineer

Related on AIssential

• The Last Private Thought — Digital privacy is eroding by default, but structural technologies like on-device AI can restore it.
• We’re open-sourcing Perception Encoder Audiovisual (PE-AV), the technical engine that helps drive SAM Audio’s state-of-the-art audio separation. Built on our Perception Encoder model from earlier this year, PE-AV integrates audio with visual perception, ac - x.com — Google's consent dialog details cookie and data usage for service delivery, analytics, and optional personalization.
• Siri won’t be your AI girlfriend — Apple's Siri prioritizes utility and privacy over engagement-driven, sycophantic chatbot interactions.
• Building Earmark: Real-time voice AI, privacy by design, and founder lessons — Earmark transforms real-time meeting conversations into actionable work artifacts, enhancing productivity and reducing manual follow-up.
• Building trust in the AI era with privacy-led UX — Privacy-led UX builds consumer trust and supports business growth by treating data consent as an ongoing relationship.
• Inside Spotify’s 2025 Wrapped Archive: AI Narratives at Scale and the Privacy Trade‑Off — Narrative-driven usage summaries enhance user engagement but amplify privacy concerns over extensive data collection.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by AssemblyAI.