I do marketing at a very small company with a hosted memory API for AI agents. Mem0 and Zep both started out open source, so a dev could clone them and run them locally without talking to anyone. Our engine is closed source, so we don't have that door, and I don't know what to use instead.
What I'd like to know
I'm not looking for growth hacks. I'd rather hear it from the dev side. If you can, answer any of these:
- Where did you first see it? When you started using a memory tool, or any tool you'd never heard of, where did you first see it? (GitHub, a post, a friend, a comparison, a tutorial, something else)
- What made you try it? What made you actually try it instead of bookmarking it?
- If you build your own: If you build your own memory and never used any of them, what did you look at before you decided to build?
- If it was closed source: If you ever started using a closed source tool, what got you in the door?
Even a one-line answer helps.
Top comments (4)
For me the first touchpoint is usually a post that shows one concrete before and after, not a landing page. Your memory example stuck with me because it names a real pain, the same question asked twice, so I would lead with that example everywhere and link it straight to a runnable demo. If the name Wontopos does not say memory on its own, let the example do that job in the first line.
Thanks, this helps. I think you're right. We explained what the product is, but we never showed one scene where it matters. I'll try opening with the example of asking the same question twice, and look for a demo people can try right away. You're also right that our name doesn't say memory, so the first line has to.
You can't open the engine, but you can open the test: publish a benchmark harness as a public repo that runs your API, Mem0 and Zep side by side with a free key. That's your version of the clone-and-run door. A demo shows yours works; a harness lets a dev compare all three on their own data.
This category already argues in benchmarks. Mem0's 2025 paper reported LoCoMo scores against Zep, and Zep answered with a post disputing how Mem0 had set Zep up. A dev picking a memory layer has likely seen that fight, so a number they can reproduce beats any claim on your landing page.
It also decides where they find you. Someone shopping here searches "Mem0 alternatives" or "Mem0 vs Zep". Run both in Google and in ChatGPT and note which pages get cited: that's the list your benchmark write-up has to land on or outrank.
One condition: the free key has to be self-serve on signup, with no approval step, or the door shuts again.
Thanks, this is a useful idea. We already have a public benchmark, but each team runs its own system on it. If we set up other products ourselves, the same fight about who configured what could start again, like with Mem0 and Zep. What we haven't done is make a ready-made door that makes it easy for others to run it. I'll talk about that with our engineer, and I'll try those "Mem0 alternatives" searches to see which pages come up.