On Evaluating Memory for AI Agents There’s been a real surge in “memory” for AI agents lately. Most frameworks now prioritize persistent memory as a core feature. But how we actually evaluate that memory is still an open question. Because memory isn’t just about storing and

Hot take: Bigger context windows are not solving the memory problem. Memory isn’t just recall. It’s: • Remembering what matters • Forgetting noise • Updating as things change

The memory layer will be bigger than the model layer. Models will commoditize. Memory won't. In five years, the difference between GPT, Claude, and Llama will be marginal. They'll all be fast, cheap, and capable. The differentiator won't be the model you use. It will be what

We believe your next favorite human...isn’t. Today we’re Born: announcing a $15M Series A to reimagine AI friends. Here's our story:

I reverse-engineered Interaction's Poke and figured out why it feels so magical. Introducing OpenPoke —an open-source implementation you can run locally. 🧵 on the multi-agent architecture and memory:

Bigger context windows can't eliminate the need for memory... A context window defines how much information a model can “see” at once. Expanding it to 200k or even a million tokens helps with short-term recall but it doesn’t help the model remember. Once the window slides, the

Discover mem0 with Taranjeet Singh as he talks about building a memory layer for AI agents and the future of personalized AI experiences. Watch here: youtu.be/O6eEcv8PR4c #AI #LLM #Mem0 #SiliconMindsHumanHearts

Yesterday, I joined TBPN to talk about what it really means to give AI a memory. And, why that’s becoming the next big moat. We discussed: - Why today’s AI apps are stateless - How memory is quickly becoming the real moat for AI systems - How Mem0 is decoupling memory from

100 billion tokens processed through OpenAI. Not a vanity metric but a signal of how fast developers are adopting the universal memory layer for AI. What matters more is what it represents, thousands of systems now learning, adapting, and remembering through mem0. Onward🚀

Applied to YC 7 times before finally getting in. Here’s what I learned the hard way👇 When I first started applying to Y Combinator, I treated the application like a pitch deck, trying to convince YC to invest, instead of explaining why the problem mattered. Over time (and 7

We’re #hiring a Data Analyst at Mem0 You’ll build our analytics foundation from 0→1 - defining metrics that matter, setting up pipelines (Postgres, PostHog, LLM outputs), and turning data into insights that guide every decision. - Bangalore, India - Full ownership - High