Replay is a provider of real-world context for the AI agent stack. For an AI agent to truly act as a personal assistant, it requires a continuous stream of context about the user's offline life—what they said, what they heard, and what they committed to. Replay's audio-based glasses act as the ingest layer for this data, converting the physical world into the digital format (transcripts and embeddings) that agents need to operate.
In the broader ecosystem, Replay is active at the hardware and data collection levels. It is building the persistent "sensory" apparatus that feeds the long-term memory of a personal agent. While many agents are currently limited to digital environments like email or Slack, Replay enables these agents to bridge the gap into physical interactions, making it a critical component for anyone building a comprehensive, 24/7 personal AI companion.
Replay is a consumer hardware company building audio-first smart glasses designed to record and index the physical world. The fundamental premise of the company is that the most valuable data in a person's life is often lost to memory—conversations, casual meetings, and spoken ideas. By housing microphones in a glasses form factor, Replay attempts to create a hands-free, passive capture loop that feeds a private, searchable database of a user's history.
The development of Replay is structured around aggressive, public milestones. Founder Zyad Elgohary completed the Mark 1 prototype in late March 2024, which served as a proof of concept for the capture-to-search loop. The current focus is Mark 2, which transitions the technology into a lightweight, usable set of glasses intended for a broader group of users beyond the founder. This hardware-centric approach is a departure from many AI productivity tools that rely on existing smartphone or laptop hardware. Replay is building a dedicated endpoint for audio ingestion.
The software layer of Replay is the differentiator. Recording audio is a solved problem, but making hours of daily conversation useful is not. Replay uses AI to transcribe audio and likely employs vector embeddings to make that text semantically searchable. Instead of remembering the exact date of a conversation, a user can ask the system about a specific topic or a promise made during a meeting. This creates a retrieval-augmented experience for the user's own life, effectively providing a long-term memory layer that is not subject to the typical human forgetting curve.
Compared to general smart glasses like the Meta Ray-Bans, which emphasize photography and social sharing, Replay is focused on personal utility. It is not trying to be a camera or a notification center. The hardware is a specialized tool for people who value knowledge management and need to recall precise details from their interactions. This puts them in the same category as companies like Limitless or the "Friend" pendant, but the glasses form factor offers superior audio positioning and less friction than a necklace or a pocket-based recorder.
Replay is currently in an early, iterative phase, taking pre-orders via Stripe as it moves toward its Mark 3 milestone, which targets general availability. The company operates with high transparency, providing daily updates and pictures of the development process through Elgohary's social channels. This transparency helps mitigate the typical skepticism associated with early-stage AI hardware, a sector that has seen both significant hype and high-profile failures.
By focusing on a single, clear utility—making life searchable—Replay avoids the feature creep that often plagues wearable technology. The success of the device will likely depend on its battery life and the privacy controls it offers, as persistent recording remains a sensitive topic. However, for a subset of power users and knowledge workers, the promise of a perfect memory is a compelling enough value proposition to justify another device on the face.
Audio-based glasses that capture and index life's conversations and memories.