Reyn Labs

Everything you've ever worked on instantly searchable

7 followers

Everything you've ever worked on instantly searchable

7 followers

Reyn is a privacy-first desktop AI for Mac that watches your screen, builds a private journal of your work, and answers questions about your week. Ask anything about what's on your screen right now. Every call captured; decisions, action items, what was said. A full daily recap emailed every evening. Built for people who lose the thread. Your work, finally remembered.

Free Options

Launch tags:Productivity•Notes•Artificial Intelligence

Launch Team / Built With

Viktor.comAn AI coworker that actually does the work

Promoted

OpenAI

OpenAI Day

How did GPT-5.6 change the ambition or scope of what you shipped?

Maker

GPT-5.6 pushed us to think bigger about what the memory layer could actually do. Earlier versions of Reyn were primarily a capture and retrieval tool. With GPT-5.6's reasoning depth, we realised we could go further: not just surfacing what happened, but synthesising patterns across weeks of captured context to generate genuinely useful outputs like SOPs, onboarding docs, and client briefs — automatically, from screen recordings, without the user doing anything. The quality of reasoning also changed how we approached live screen Q&A. We went from simple retrieval to real-time contextual analysis. Reyn can now look at what's on your screen and give you a thoughtful answer that accounts for everything it has learned about your work over time, not just what's visible in the current window. That combination, passive capture plus deep reasoning plus long-term context, is what turned Reyn from a smart recorder into something that feels closer to a second brain. GPT-5.6 made that leap feel bui

Report

1mo ago

Maker

📌

hey Product Hunt 👋 i'm Goziem, co-founder of Reyn. the honest version of why i built this: i was in back-to-back calls one afternoon, and by 5pm i couldn't tell you what i'd decided, what i'd promised anyone, or how the day had disappeared. i opened an AI tool to help me catch up and spent ten minutes just explaining what i was working on before i could even ask a question. that was the moment. every AI tool i used was smart, but none of them knew anything about me. i had to re-explain my project every time. i'd leave calls and things would fall through. i'd finish a day with no honest picture of what actually happened. Reyn is what i wanted to exist: something that watches what i'm working on, remembers it, and is ready to help without me briefing it every time. a lot of our early users have ADHD, which makes sense. when your brain doesn't hold context reliably between tasks, losing the thread isn't a bad habit, it's just how it works. Reyn doesn't ask you to build a system or remember to log things. it holds the thread for you. what makes it different from other tools in this space is that Reyn is active, not just a recorder. you can ask it about your screen mid-task, get a real picture of your day, and have an AI that accumulates genuine context over weeks. we're early. there's a lot still to build. i'd love to know: what's the part of your workday you most wish something was quietly paying attention to? — Goziem

Report

1mo ago

Hey PH, cofounder here too 👋. I built Reyn's capture engine and put most of my thinking into the privacy architecture, since "an AI that watches your screen" only works if people can trust exactly what happens to that data.

The short version: capture is triggered by real activity (not a continuous recording, stops when idle), and three filters run before anything is stored:
1. a pre-capture blocklist for password managers/2FA apps
2. a post-capture text filter that drops any frame matching sensitive keywords, and you can extend both, but never weaken them.
3. Screenshots are sent once to summarize into text, then the image is deleted - never stored, on-device or in the cloud. Your encryption key lives in your OS keychain, never on our servers.

On privacy: I won't say "your screen never leaves your device," because frames do transit to an AI model to get summarized (we added BYOK to provide some optional user control here). These frames are deleted right after and never stored, but they do leave during that step. On encryption: credentials are locked with your device-held key, but timeline summaries are protected by access controls + encryption in transit/at rest, not full end-to-end.

Wold love to hear from you all! Does the model of transient processing, nothing stored, but not literally zero-transit — feel like the right tradeoff to you? What would change your mind either way?

Report

1mo ago

Reviews

Wold love to hear from you all! Does the model of transient processing, nothing stored, but not literally zero-transit — feel like the right tradeoff to you? What would change your mind either way?