News for hackers. Stuff that matters.

Thoughts on memory layers like mem0

Really curious if there is actually a market for companies trying to sell a memory layer for AI agents.

Look at Claude code's "automemory" for example, they don't use a vector database or anything complex. Its just simple markdown files on the local filesystem along with grep for retrieval and it works great for me.

For many simpler use cases, simple techniques like context summarisation are good enough.

( 7 comments | on HN )
Re:Thoughts on memory layers like mem0 by gregzem on Sunday November 15, 2026 @06:29AM (#50506114)
It depends on how you productize it. There’s market. The memory itself is a must-have, but you should probably implement it with some advantages and make it domain-specific to be better at something like retrieving, indexing, etc.
Re:Thoughts on memory layers like mem0 by anshmalik2002 on Sunday November 15, 2026 @06:58AM (#50506280)
I agree that domain-specific memory layers might be the way to go, in which case devs are just better off making their own memory layer for their use case, instead of having a dependency on a third party, don't you think?

Especially now that building a simple memory layer particular to your use case won't take a lot of time and effort. And you can keep making it better as you go forward.

Re:Thoughts on memory layers like mem0 by muzani on Sunday November 15, 2026 @07:12AM (#50506365)
There's certainly a market. Heck, the chatbot market share likely reflects how good this memory is. Grok and Gemini are good products, but not known for retaining memory, and despite having the distribution, they lose out to ChatGPT.

I do think it's actually a Hard problem, otherwise everyone would have done it. ChatGPT is the king at it, but they're not selling memory as a product, so I assume it's tightly coupled with the chat. Claude can't do it without token guzzling.

And the first thing I set up with AI chatbots is memory. You need memory to work with past conversations, and suddenly you're dealing with emergent behavior and token drain.

The other question is timing.

It's probably a year late. You could easily make a billion dollar company just from this, but just as easily be disrupted by Big AI.

There will probably be a time when this would be unbundled, similar to how telcos and cloud unbundle all kinds of offerings. And that opportunity will come several years later.

Re:Thoughts on memory layers like mem0 by anshmalik2002 on Monday November 16, 2026 @07:29AM (#50516901)
Don't use grok and gemini, but based on your comment, I am surprised that they haven't cracked memory yet. I actually like Claude code for memory, but you're right from a token consumption perspective, it consumes a lot more tokens, didn't really think much of this before.

I agree that it is a hard problem at scale. But at that scale, companies should be able to build their own memory layer for their agent's use case, instead of relying on a memory layer which is built for a more general use case.

Re:Thoughts on memory layers like mem0 by muzani on Monday November 16, 2026 @08:56PM (#50526417)
I think there's space for a Heroku/Netlify/Firebase style service. Somewhere between large scale and none.

It took me a few weeks and some talented/passionate engineers to figure out embeddings and vector DBs and searches. And another few weeks to set it up to a production level. Markdowns are cute, but every second of processing time is the difference between people using it and people dropping off.

If you have something that's just plug and play, and fairly cheap to start with, that would help a lot of people. Some would just stay locked in rather than hire an engineer to deal with it.

Re:Thoughts on memory layers like mem0 by costinu on Sunday November 15, 2026 @03:04PM (#50510162)
Local md files work fine when it's just you on your own machine, but for an app/service with thousands of users, using md files will be challenging.

You also need to handle duplicate or conflicting facts, and that becomes a burden as the memory grows.

Many app builders might start using a vector database. This will help with storage and organising data, but will fall short on keeping an accurate memory when facts change.

So I think there is a market, but only for the multi-user case.

Re:Thoughts on memory layers like mem0 by andrei_gusev on Tuesday November 17, 2026 @09:36AM (#50531953)
There is a market, but for now it's all concentrated in business and teams, especially in enterprise solutions. I'm building a similar service.

When I was renewing my Japanese business visa, and that's a very bureaucratic thing, Claude's memory really let me down. It kept forgetting different numbers, mixed up dates, which documents were done and which weren't yet. I built a small project for myself, and it helped a lot. Then I rolled it out for everyone.

It's growing organically, from Google, from the ChatGPT and Claude directories, so users come with a problem and solve it. Mostly these are teams and small business owners. But when I tell people about a separate memory layer, they say their Claude or GPT handles it fine. So there is a market, but not among individual users.