We built git for AI memories: version history, editing & full data control
If I could sum up building in public in two words, it would be: course correction.
We launched AI Context Flow on Appsumo a few weeks ago, and turns out, managing AI context across tools is more nuanced than we thought. Thousands of people tried our product and one piece of feedback kept recurring: users needed more control over memory curation.
The demand was clear, but our product had missing features. So we did what any startup team does: we course corrected and went to our cave to build out what users needed.
Here's what we just shipped:
🔍 Full memory view: click any memory to see raw content, when it was added, and which source it came from.
✏️ In-place editing: edit memories directly from the UI (Notion style) or via MCP. No workarounds.
🔁 Version history: the first of its kind: a git for memories. View, compare, and restore previous versions.
📥 Download your data: local copy, anytime. Your data is yours and we mean it.
🗑️ Delete via MCP: delete memory buckets directly in chat. Previously it was UI-only; that was creating friction for chat-first users.
Try it now → https://app.plurality.network
We are also hosting a live AMA session this thursday to showcase the features shown above and answer any questions you might have. Register here: https://luma.com/g092y5pg


Replies
Love the focus on memory transparency . Users should always know what the AI remembers and why.