Mem0: Building Production-Ready AI Agents with Scalable Long-Term Memory — mentioned by elasxies

Tue, 12 Aug 2025 23:50:54 GMT

I’m so tired of seeing desperate tech startups lie on their numbers and slap SOTA for VCs. Mem0 published this paper: https://arxiv.org/abs/2504.19413, claiming they beat everyone on LOCOMO, but turns out they completely botched the implementation of their competitors. Then used those bad results to prove their dominance. When Letta and Zep actually ran the benchmarks correctly, they both hit a 10% higher score than Mem0’s best. There's so much vaporware in this industry it's insane. I get the incentive to get VC money by overpromising features, but lying on research papers is pathetic. do…

Paper and discussion history. A mention is not an endorsement.

Mem0: Building Production-Ready AI Agents with Scalable Long-Term Memory · HackerPapers discussions

Mem0: Building Production-Ready AI Agents with Scalable Long-Term Memory — mentioned by elasxies