Ask HN: How Valuable Is Plug-and-Play Scoped Memory for AI SaaS Apps?

[designorbit]

Hey HN, I’m working on a new tool called Recallio that aims to solve a problem I’ve personally hit while building AI-powered apps: LLMs (like GPT) are great, but they forget everything between sessions unless you custom-build a full memory layer.

We’re testing an API-first service that lets you:

- Scope memory per user, agent, or project - Write + recall data instantly via clean APIs - Set TTLs (auto-expiration) and export your data anytime (no lock-in) - Skip the infrastructure hassle of building your own DB + logic

We’re not trying to be an agent framework—just the memory brain you can easily plug in, so your AI can remember what matters.

I’m curious: - How are you solving memory for your AI agents or SaaS tools? - Do you rely on vector DBs like Pinecone/Faiss, or hack together custom solutions? - What’s missing in current solutions that’s been painful?

We’re in early access mode and would love feedback, ideas, or war stories. Here’s the landing page if curious: [recallio.ai]

Thanks in advance for any thoughts Andrew @ Recallio

[bfeynman]

This doesn't tell me anything about how you actually solve the problem in a value-add way. There's no technological insight? Are you just putting this behind an API (adding more latency). The thing about memory is for non academic cases the value of it is still nebulous and hard to quantify, another reason why it still feels experimental, and why I would find it hard sell to push that to a black box service. There's nothing on your website about actual performance...