Contextberg
Product Hunt Tiger
Contextberg is a local memory app for AI agents on Mac and Windows. It keeps screens and chats private on-device, then shares only the bits an agent needs.
Based on reporting by Product Hunt, Tiger — read the original for the full story.
Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error
Contextberg is pitching itself as memory for AI agents, but the twist is where that memory lives: on your own Mac or Windows machine. The app records screens, browser history, and agent conversations, then turns all of it into a private archive that can be searched later and fed back to tools like Codex, Claude Code, and Cursor through MCP.
That matters because agent workflows get messy fast. A model can only answer well if it can see what happened before, and Contextberg is trying to make that context easy to pull up without dumping your whole workflow into someone else’s system. The archive stays on-device, while only the selected context goes to the provider you choose.
This launch adds native macOS capture, OCR search for screenshots, source exclusions, and more routing options. Users can lean on an existing Codex sign-in, a Gemini or OpenRouter API key, Contextberg Cloud, or a fully local model. That’s a lot of ways to wire it up, which is basically the point.
It’s also the product’s second launch on Product Hunt, so this is clearly still early days. But the pitch is straightforward: keep the memory private, make it searchable, and hand agents only the useful pieces instead of the whole mess.
My take — AI-written commentary, not fact-checked reporting
This is the right kind of boring. Private by default, selective when needed, and not pretending every prompt deserves a trip to the cloud. The industry keeps acting like “AI memory” has to mean bigger and leakier; Contextberg is making the opposite argument, which is refreshing and slightly overdue.
Read more about this at: Product Hunt