cognee + G.V() → classics like Moby-Dick become queryable memory graphs (NL or Cypher) you can see instantly. Big thanks to @gdotv.bsky.social. Read the full write-up from the below!

Memory Monday: Graph It Out! Who’s working with memory graphs this week? Share what you’re connecting - entities, intents, sessions - whatever! We love those node clusters

Here’s the distilled version of what everyone has been talking about “context engineering”. Straight talk on memory layers, GraphRAG and why prompt hacks alone won’t cut it. Give it a read 👉🏼 dub.sh/context_engi...

We’ve been jamming on context engineering for weeks. Pulled everything we’ve seen—memory, RAG, graphs, cost—into one post. Give it a read 👉🏼 dub.sh/context_engi...

🧠 Launching AI Memory Monthly in a few hours! One ultra-skimmable email → fresh papers, real-world retrieval tricks & the spiciest AI memory debates. Hit the subscribe button now so Issue #1 lands in your inbox soon.

We’re debating everything from RAG pipelines to memory compression in a spicy SubReddit. Add your hot take! 🔥 Then swing by our Office Hours on Discord today (17:00 CET) to keep the convo rolling.

Hello, world—with context! 🧠 🚀 We are launching “Insights into AI Memory”—a monthly signal on the tech, tools & people teaching AI to remember. Grab the initial post & subscribe 👉 aimemory.substack.com

We’re days away from opening the cognee SaaS beta 🚀 where you’ll bring your knowledge graphs & LLM workflows to life without the infra pain. 🔍 Built for everyone who cares about clean data, speed, and reproducibility. Want in on day 1? Join the waitlist → dub.sh/beta-saas-co...

Woke up to 🚀 cognee hitting 5000 stars! Thank you for the trust, feedback and code you pour in. Let’s keep building 🛠️

find us on @github.com trending!

Yesterday, we released our paper, "Optimizing the Interface Between Knowledge Graphs and LLMs for Complex Reasoning" We have developed a new tool to enable AI memory optimization that considerably improve AI memory accuracy for AI Apps and Agents. Let’s dive into the details of our work 📚

Join the conversation at r/AIMemory. dub.sh/ai-memory

Full write-up → www.cognee.ai/blog/fundame... If you’re exploring how to blend vectors and graphs for richer retrieval, we build exactly that at @cognee.bsky.social - DMs open for a chat!

We just dropped a follow-up to last week’s graph-DB explainer—this one dives into vector databases and why they’re the workhorse behind semantic search. Quick recap: Graph DB → “How are things connected?” Vector DB → “What *feels* like this thing?” 🙂

👉 Ready to deploy? ``` git clone <https://github.com/topoteretes/cognee.git> cd cognee/cognee-mcp brew install uv uv sync --dev --all-extras --reinstall source .venv/bin/activate ``` Full docs: docs.cognee.ai/how-to-guides/deployment/mcp

🚨 4 Big Updates to cognee MCP Server (and what devs need to know): If you're building with LLMs, graphs, or agentic applications - keep reading. 👇

and more… Our latest post, “Graph Databases Explained,” walks through the trade-offs, query patterns, and performance internals in plain language. If “deeply connected data” describes your next project, it might save you a few JOIN headaches. Read here → www.cognee.ai/blog/fundame...

Got featured in @neo4j.com ’s weekly wrap-up! From the links below, you can find us chatting on GraphRAG + cognitive science (a small demo included) + fresh Neo4j news (Aura Analytics, NODES CFP, MCP hacks). Don’t miss it out.

🧵 Insights from Data Council 2025, viewed through my lens:

Curious if small Ollama-hosted LLMs (e.g., Phi-4, Gemma-3) can reliably create knowledge graphs? Our 3-minute video answers exactly that: - small local model limitations - tips (schema optimization, YAML/HTML requests) - need for larger models (Llama3.3, 14B) watch here: youtu.be/P2ZaSnnl7z0

We benchmarked cognee, Mem0 and Zep/Graphiti using HotPotQA! Cognee topped the charts, boosted further by Dreamify, our optimized pipeline 🚀. Graphiti shared their scores with us and we'll verify them soon! Dive in the details 👇🏼

Join us at @pyconde.bsky.social in Darmstadt - this is your opportunity to dive into open-source, network with passionate devs, and shape the future of AI and Python ecosystems! We are sprinting with @cognee.bsky.social alongside 9 other awesome open-source projects.

TLDR; we launched on ProductHunt! After months packed with community-driven features our AI memory tool @cognee.bsky.social has evolved to new heights. Finally AI agents meet the memory they deserve - structured, accurate, and reliable - in 5 lines of code.

Exciting news friends! Our latest @cognee.bsky.social update is live, featuring: • Ontology support • Direct relational DB imports • Dreamify, our shiny new hyperparameter tuning system • Big performance boosts Plus, Pinki our dog has finally conquered the indoor puke challenge.

Give LLM your rules? Your definition of a session, GBV or the details of your accounting practices. Yes. With ontologies! Connecting related concepts across different documents is often necessary but traditional search methods can’t help since they treat each paper as an isolated document.

Short read, big insights 👇 We took knowledge graphs up a notch with formal ontologies. You’ll find how we merge domain expertise + graph structures for more detailed insights at @cognee.bsky.social in the below blog. Check it out.

How do you run your own local AI stack? In 4 minutes, you can setup Ollama with @cognee.bsky.social to go local with Phi-4 and Mistral Going local mostly means: - Better privacy - Reduced costs - More control & experimentation But.. beware:

🧵 7 Common Misconceptions About Knowledge Graphs (and what devs should actually know) If you’re working with data, LLMs, or building anything agentic —you’ll want to read this.

Want your AI to catch sneaky data errors and go beyond basic vector search? We just released a blog on using knowledge graphs for smarter QA. Check it out! ps: a sneak peek from the blog

What’s the deal with Knowledge Graphs? TL;DR: They don’t just store facts. They connect them. They power AI memory, enable deep reasoning, and help LLMs process data with context - but what are they, really?

Everyone’s hyped about MCP now, but cognee was already there. It hooks your LLM into your data - text, code, you name it - so you can cut hallucinations and keep context. I use it with Anthropic’s Claude Desktop, plus dev tools like Cursor, Cline, Continue and the difference is huge.

Let's turn your data into graph/vector representation! We will cognify it. Inspired by #graphrag and #rag, we have a better approach! Why? - Increase retrieval accuracy by 30% (measured on hotpotqa) - Get the data you need - Reduce the number of LLM calls - Extract the data from your own data

Just dropped our second video on: - understanding graphs & knowledge graphs - using cognee as a unified layer between your data & AI - building knowledge graphs with it - deploying with Modal Check it out: youtu.be/86SWVdI5K0Y

🎥 Check out our new video tutorial on integrating cognee with Claude: www.youtube.com/watch?v=fI4h...

RAG evals shouldn't be a guessing game. With DeepEval + cognee, you get clear, structured insights into how your retrievals are performing. No more blind spots.

🚀 Open Source Synergy Calling all OSS developers! I am happy to announce a seamless integration between cognee and Milvus: 🔗 Transform raw data into structured vectors and knowledge graphs. 📊 Deliver accurate, context-aware responses. ⚡ Scale effortlessly with fully open-source pipelines.

If you like to read about AI memory and how we approach to it, here is our blog post: www.cognee.ai/blog/fundame...