What should AI agents remember beyond facts? We're building Kumiho Memory.

Hi Product Hunt! We're the team behind Kumiho Memory, launching here in a few hours.

We built Kumiho around a problem we kept running into: an agent might remember what we decided, but lose why we decided it, which alternatives we rejected, or what changed later.

Our approach is a shared, versioned memory graph across agent tools. Decision Memory keeps rationale and rejected alternatives alongside decisions; code_why connects code changes to their rationale. The aim is continuity across sessions and tools, not just more retrieved text.

Before launch, we'd love to hear from people building or using AI agents:

- Where do your agents lose context today: between sessions, between tools, or between teammates?

- When a past decision changes, what should be updated automatically, and what should require your review?

- What would make you trust a recalled memory enough to act on it?

Tell us about a real failure case. Your feedback will help us prioritize what to improve next.

Explore Kumiho: https://kumiho.io