Prashast Vaidya

[mem0 @mem0ai]

Mem0 is now available on the Vercel Marketplace!

[mem0 @mem0ai]

Introducing Gateway.

Your agent needs to check a payment. You hand it a key that can also refund it.

It needs to draft a page. That key can archive every page.
Not anymore.

Gateway gives every agent one key with only the tools you grant. MCP servers and your own APIs, same key. Change the grant when the job changes. The agent never sees your real credentials.

Beta customers already run over 95,000 calls each through it.

Watch it work ↓

[Deshraj Yadav @deshrajdry]

We built Mem0 Gateway since giving an agent a task shouldn’t mean handing it every permission your API key has.

Check it out👇

[mem0 @mem0ai]

Introducing the token-efficient memory algorithm.

Mem0’s new algorithm hits 91.6 on LoCoMo, 93.4 on LongMemEval, and 64.1/48.6 on BEAM (1M/10M) while averaging < 7,000 tokens per retrieval call, at 3-4x lower token cost. Full-context approaches on these benchmarks run at 25,000+ tokens.

What it means for developers:
You can now run agent memory that is cost-efficient at production scale without sacrificing quality.

What’s changed:
- Single-pass ADD-only extraction
The previous system used two LLM passes: identify candidates, then reconcile with ADD/UPDATE/DELETE. That reconciliation destroyed context. The new algorithm: one pass, only adds. When information changes, both facts survive. The system reasons about how things evolved, not just where they landed. ~50% faster extraction.

- Agent-generated facts are first-class
Users and agents both produce facts during a conversation. Now both are stored with equal weight. This change increased the single-session (assistant) score from 46.4 to 100.0 on LongMemEval.

- Entity linking
Every memory is analyzed for entities, people, places, concepts. These are embedded and stored in a separate lookup layer, linking memories about the same person, place, or concept across sessions and months.

- Multi-signal retrieval
Retrieval stack now runs three scoring passes in parallel and fuses the results: semantic similarity, keyword matching, and entity matching. The combined score outperforms individual signal scores on every benchmark.

- Keyword normalization, Queries like “what meetings did I attend?” now match memories containing “attending a meeting.” Small change, measurable impact.

Available today on platform + open-source SDK.
Full eval framework open-sourced.

Full writeup and benchmarks in thread 🧵👇

[Deshraj Yadav @deshrajdry]

We launched OpenMemory MCP. Check out the announcement 👇

https://t.co/OQsh1ybXNg

[Taranjeet @taranjeetio]

We’re excited to launch OpenMemory MCP, a private memory for MCP-compatible clients powered by @mem0ai

Today, most AI assistants and dev tools operate without memory. You plan your roadmap in Claude, implement tasks in Cursor, but none of them know what the other did. Each tool operates in isolation, and your context disappears as soon as the session ends.

OpenMemory MCP solves this.

Built on the open Model Context Protocol (MCP), OpenMemory MCP runs 100% locally and provides a persistent, portable memory layer for all your AI tools. It enables agents and assistants to read from and write to a shared memory, securely and privately.

Key capabilities:

✅Works across MCP clients like Cursor, Claude Desktop, Windsurf, Cline and more
✅Provides standardized memory operations (add_memories, search_memory, list_memories, delete_all_memories)
✅Stores data on your machine, fully private
✅Offers a centralized dashboard for visibility and control
✅Simple Docker-based setup with no vendor lock-in

If you're building on MCP, OpenMemory MCP is the easiest way to add persistent, private memory to your clients with zero cloud dependencies and full local control.

We’ve put together a short demo to show how it works

Check it out in the links below 👇

[Prashast Vaidya @PrashastVaidya]

Mem0 setting industry standards for fast and eidetic agents! 🚀🚀

[Taranjeet @taranjeetio]

Today, I’m excited to share our latest research on “Building production ready AI Agents with Scalable Long-Term Memory”.

We’ve achieved state-of-the-art (SOTA) performance—26% more accurate than OpenAI Memory.

We evaluated Mem0 on the LOCOMO benchmark and found that it consistently outperformed all the six baselines on all types of questions from multi-hop to temporal. Mem0 reduces latency by 91% and cuts token usage by over 90% compared to full-context methods, offering fast and cost-effective performance.

Today’s AI agents quickly forget important information once it moves beyond their context window, leading to broken conversations, repeated mistakes, and lost user trust. Larger context windows only delay the issue - making systems slower, more expensive, and harder to scale.

Mem0 was built to solve this head on - giving AI Agents a scalable memory layer that remembers what matters, reasons faster and adapts over time.

Check out the full paper below 👇🏻

[Taranjeet @taranjeetio]

We launched @mem0ai on YC &amp; a lot of users reached out to learn more about what we’re building.

We have written a blog which describes what Mem0 is &amp; why we believe memory will be very core for the upcoming agentic world.

Link in next tweet 👇🏻

[Deshraj Yadav @deshrajdry]

🚀 Launching Claude Memory: Long-term memory for Claude

💡 I recently switched from ChatGPT to @AnthropicAI but miss OpenAI's memory feature. So, I built the missing piece with @mem0ai in 4 hours.

🧠 Problem solved. Try it now!

🔗 Link in reply

[Taranjeet @taranjeetio]

Open sourcing Dynamic Graph Memory by @mem0ai.

Memory is fundamental to human reasoning, shaping how we approach tasks and make decisions. At Mem0, we believe that AI agents & apps should reflect this principle.

Our Dynamic Graph Memory emulates human memory, advancing AI agents toward more intelligent, human-like reasoning.

This is a significant step forward in building AI that truly understands and interacts with the world like we do.

All credit to @khant_dev @deshrajdry @pckraftwerk for their countless nights spent on bringing this to life.

Link: https://t.co/lWaBf6Gj8q

[Taranjeet @taranjeetio]

Introducing memory in @embedchain - personalized, reliable, and adaptive memory for LLM.

Features:
• Works similar to a human brain
• Knows what’s important, why it is, when it is, and how much
• Helps in reflection and planning activities
• LLM agnostic

👇🏻👇🏻👇🏻

[Taranjeet @taranjeetio]

Introducing @embedchain - the easiest way to manage data for LLMs to tailor them to your own use case.

Think of it as the “Data platform for LLMs”

https://t.co/em2UdY2cjG

[Taranjeet @taranjeetio]

🚀 We've hit some big milestones with @embedchain:

• 300K apps

• 53K downloads

• 5.7K GitHub stars

Every step taught us something new.

Today, we're taking those lessons &amp; introducing a platform to manage data for LLM apps.

No waitlist, link in next tweet 👇🏻👇🏻👇🏻

[Taranjeet @taranjeetio]

👀👀👀

coming soon...

@embedchain

[Prashast Vaidya @PrashastVaidya]

@deepakorani0 @WarshaKorani Congratulations Warsha🎉

[Prashast Vaidya @PrashastVaidya]

@saahil_parikh 🙄 footage khaane aaya Hain yeh

[Prashast Vaidya @PrashastVaidya]

Same ball! Same opponents! I don't understand why SKY should be dropped! Are the senior players that good? I know they are legends of the game but shouldn't the current form be a factor?

#INDvSL

[Prashast Vaidya @PrashastVaidya]

@AskAmex I have the Amex Platinum card and was booked through Amex Travel online website.

[Prashast Vaidya @PrashastVaidya]

Hi @AskAmex I've been trying to get my flight changed after you changed my itinerary in an extremely poorly designed email. Been trying to reach a supervisor via phone but to no avail. Please help with better customer service!!

[Prashast Vaidya @PrashastVaidya]

So 3 hrs up! I cut the call and they call me back in 1 min and tell me can you hold! Why can't they do that when I am on hold! This is the most frustrating moment of my life and the worst customer service experience I have ever received!

[Prashast Vaidya @PrashastVaidya]

This is ridiculous! I have been on this customer service call for almost 2 and a half hours and have been on hold for the last 30-40 mins!