Home< BlogCognee News
Apr 3, 2025
4 minutes read

🚀 cognee Update: March 2025

Vasilije Markovic
Vasilije MarkovicCo-Founder / CEO

Hey there, cognee community! 👋

We're thrilled to share another round of exciting updates with you! In our latest sprint (we're averaging 3-4 releases per month now!), we've been cooking up something special that we think you'll love.

Before we dive in, here's a fun fact: Our team has pushed over 100 commits this month alone, and we've gotten +400 Github stars!** 🎉

On a lighter note, this month marked two firsts at cognee HQ: our office dog, Pinki, finally made it through without puking indoors, and we're rapidly approaching the infamous PR #666.

What's New?

✨ Ontology Support is Here!

You asked for ontology support—and now it's here, thanks to your valuable feedback (big shoutout to our awesome Discord community)! But what exactly is an ontology?

An ontology is a formal representation of knowledge that defines:

  • Concepts within a domain

  • Relationships between concepts

  • Properties and attributes

You can now effortlessly add your OWL files, and they'll be matched automagically! See the details in our docs.

Why use ontologies? They help you ground LLMs with facts. Have a look at our example with German car makers.

Here's how smooth it looks in practice:

🎯 Import Your Relational Databases Directly

Say goodbye to manual database conversions! We've added relational database import capabilities, making it seamless for teams to bring structured relational data into cognee graphs. Why is this useful? We can store the data into the graphs and import your business metrics and let AI Agents use them. Find out more here (link to post)

🔧 Dreamify: Our Hyperparamagician

Rounding out our updates is Dreamify, a tool that has been in high demand among our customers. Dreamify—our hyperparameter tuning system—is here to turbocharge cognee’s performance.

Dreamify delivers:

  • Enhanced prompt management
  • Improved evaluation systems
  • Significantly faster performance

⏱️ Performance Improvements

We're obsessed with keeping our codebase tidy, and it shows! This month alone, we’ve:

  • ⚡️ purged 85 unnecessary files
  • 🎮 implemented 10 new tests
  • 📈 dramatically improved graph performance

Getting Started

Ready to upgrade? It's quick and painless (takes under 30 seconds):

Then:

  1. Check out our docs at docs.cognee.ai
  2. Start building amazing things!

What's Coming Up Next?

We're already cooking up the next batch of improvements (our roadmap is packed with goodies!):

  • 🎨 A brand-new, user-friendly UI you'll love to use
  • 🤖 Cognitive process mapping and tools for managing controlled hallucinations
  • 🌐 Massive scalability improvements, enabling cognee to run smoothly across hundreds of containers simultaneously

Community Corner

Your feedback shapes every feature we build. Want to get involved?

  • 🎮 Jump into our lively Discord chat to learn how to build your own AI memory, get a say in what future cognee updates bring, and to share & learn from our user hivemind!
  • 💻 Contribute to our codebase on GitHub to help shape cognee firsthand, learn new things, and become a recognized member of our growing open-source community.

Graphing Our Progress

This update is a big one! We've packed it with improvements that make cognee even more powerful and user-friendly. As always, we're committed to making your development experience better with every release.

Keep an eye on our GitHub releases for more updates, and happy coding! 🚀

Get started

Cognee is the fastest way to start building reliable Al agent memory.

Cognee Cloud
Latest
What Is a Knowledge Base? (and Why Most of Them Stop Working)
A knowledge base is a centralized system for storing reusable information — but most fail because of ownership gaps, drift, and no clear sense of what actually belongs in them.
LLM vs Generative AI: Comparing Models, Memory, and Architecture
Generative AI and LLMs are not the same thing. Learn the real difference, why architecture matters more than model size, and what memory and retrieval actually do.
Best Vector Database: Choosing for Search, RAG, and AI Memory
There's no single best vector database — the right choice depends on your retrieval workload, deployment model, and whether you need search, RAG, or full AI memory.