Unabyss for Claude - Shared memory across all apps and LLMs. In Claude
by•
Claude doesn't know what happens in GPT. Neither one really knows who you are or what your company does. Now they can.
Unabyss gives Claude memories from your other AI agents and everyday apps: email, Drive, GitHub, Notion, meeting recorders, and 20+ more. It saves new memories too, so GPT and Cursor stay in sync with the exact same context - sharper than wiring each tool into Claude one by one.
Finally, a real memory that follows you. Private. Portable.
Replies
Philip, having to explain who I am all over again every time I start fresh gets old fast, so this really clicks for me. Keeping a hand on what each tool gets to see is the part I appreciate most.
Unabyss
@matthieu_poitrimolt thanks!
Kalendar.work
Really like the direction. Curious what the initial setup looks like time-wise, is it minutes to connect your core apps, or more of a gradual thing as it learns you? Either way, nice launch!
Unabyss
@aima Thanks, so the initial setup takes up to one and a half minutes. Then, depending on how many items are pulled from connected apps, the first sync will take minutes; larger syncs take 10-30 minutes. Either way, it's a one-time setup, and after a few minutes you can start getting value from Unabyss, and that value only grows the longer you use it.
Finally something that fixes the endless copy-pasting of context into every new chat. Hooked it up to my notes and Linear, and Claude actually remembered what I was working on without me prompting.
Unabyss
@celalgmubusiuw awesome!
Epic! Looking forward to trying this out.
Congrats on the launch 🚀
Unabyss
@michael_yagudaev thanks!
HarnessRouter
@malpunek Congratulations. And happy product launch.
Unabyss
@huisong_li Thanks a lot!
The MCP-first approach is the right call, and I like that conflicts get flagged to the user instead of the system silently picking one source over another. Most memory tools quietly decide what's true about you, this one at least asks.
One thing I didn't see addressed yet in the thread, when context gets pulled from something like Slack or Gmail that includes other people's names and info alongside yours, is there any filtering to keep that out of your personal vault, or does the extraction just take whatever's in the source as-is.
The Eureka Database
Congrats on the second launch, Philip and team! I’ve definitely burned way too many tokens re-explaining the same product, preferences, and project context every time I move between Claude, Cursor, and ChatGPT, so this immediately feels useful.
The only thing I’d worry about is context getting bloated over time. Would love a simple way to see exactly what memory is being pulled in, how many tokens it uses, and quickly pin, trim, or exclude certain context before it gets passed to another tool.
Unabyss
@jeremy_galang thanks for the comment & feedback!
When it comes to the problems you described - we're actually working on showing exactly what's in your memory in a clear way. Token usage - that's interesting and we're currently researching that - just to check how much tokens are saved thanks to Unabyss. I believe that eventually you save them rather than burn tokens with additional context delivered via MCP.
Finally something that fixes the “let me paste my whole life story into the prompt again” problem. Hooked it up to my notes and Gmail, and Claude actually remembered my project context without me retyping anything.
Unabyss
@asiyel42174 we are glad you like it!
I run a hand-rolled version of this for a fleet of Claude agents — plain files, one fact per file — and the failure mode that taught me the most wasn't retrieval, it was propagation: one agent writes a fact that's slightly wrong or goes stale, and every other agent confidently inherits it. So my question is about contradictions: when a fresh observation from Gmail disagrees with an old memory that came from Notion, does the old one get overwritten, versioned, or decayed? And can I audit which app wrote a given memory? Provenance is the part I'd actually pay for.
Unabyss
@mystoryland Hi Olga, in Unabyss, you can see what the source origin is. In terms of conflicts, we have a resolution engine that checks incoming memories against existing data and evaluates them based on date, origin, author, and other factors. Most of the time, it handles it well, but if it gets unsure, then it will leave a note next to the change, explaining why and when it happened.
@dominik_bartosik The "leaves a note when it's unsure" part is the answer I was hoping for — silent auto-merge is exactly how a wrong fact propagates everywhere before you notice. Surfacing the low-confidence resolutions instead of hiding them is the honest call. Nicely done, and congrats on the relaunch.
Congrats on your relaunch!! Have you seen more adoption from individual power users or teams?
Unabyss
@tarqiya_forgah Thanks! We have a lot of power users, but interest from teams is strong too. We will be launching team plans next week to accommodate the interest ;)