Vasilije
@tricalt
Followers
1K
Following
575
Media
194
Statuses
2K
https://t.co/uFkKaD1CSH + Big Data | Vizsla | Pizza oven | Psychology | Data Eng. Community👇
Berlin
Joined February 2014
Yesterday, we released our paper, "Optimizing the Interface Between Knowledge Graphs and LLMs" 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 📚
1
6
33
Trying to make our content about agents and memory more approachable to users led us to some South Park level ideas
0
0
0
Grateful to our community and @boundaryML team for all the feedback and support! Read more here:
cognee.ai
Elevate AI memory with cognee + BAML: type-safe LLM outputs powering memory for AI agents in production. Validate schemas, cut errors at scale—start building now!
0
0
1
BAML gives you predictable outputs, simpler error handling, and a smoother path to production for LLM features cognee dynamically transforms your Pydantic models into BAML types and provides a single unified function for structured LLM outputs. Developers benefit from: -
1
0
1
We kept hearing our users asking for stronger guarantees around structured outputs and less prompt drift as projects grow. So we built BAML (by @boundaryML) into @cognee_
1
1
4
🧠LangGraph × cognee Integration cognee brings persistent memory to LangGraph agents, letting AI applications maintain context across sessions while seamlessly working with existing LangGraph features. Check out how to add memory to your agents 🔗 https://t.co/Om6FZE2gRR
7
35
238
Don’t build Shallow Agents on LangGraph. Let’s get deep. Deep Agents require persistent memory. With two simple tools from our new integration, you can now - add any data to @cognee_’s semantic memory - inspect the knowledge optionally and - let your agent retrieve what
1
3
16
Neo4j × cognee is a great fit We made “This Week in @neo4j ” again. This time with our self-tuning AI memory. You can read how we built memory that learns from real user feedback. Thanks @alexandererdl Read here:
1
2
9
SF âś… London âś… Same red stage, new questions. Spoke at #RedisReleased2025 on Agent Context w/ @basetenco & @tavilyai Big thanks to the stellar @Redisinc crew!
0
5
12
Yesterday we closed our very first Contribute-to-Win challenge. Heart full. We have 7 winners! We’ll split the prize evenly + swag for all seven. Huge thanks to first-time-ever contributors, cognee newbies, and old friends. What should be the next challenge?
Contribute-to-Win wrapped! We finished final reviews yesterday as we shipped cognee v0.3.5. The leaderboard ends in a 7-way tie—so we’re splitting the prize pool equally and sending swag to all 🎉 Thank you first-timers and familiar faces! What should be the next challenge?
0
2
8
Before pushing forward, let's take a quick step back. Proud of the @cognee_ team and community: - the positive and constructive feedback on all our product announcements, - watching builders put the new cognee UI through its paces, - the quality of conversations at Redis
1
1
9
Read the full-write up from the link below and learn how user management system works in cognee.
1
0
2
Here is a tutorial on how to get your AI agent reason about Python like @gvanrossum Build a semantic memory with @cognee_ by unifying multiple sources of truth (including Guido's PRs and commits, developer chat history, rules & docs) into a queryable knowledge graph backed my
Building AI agents that can synthesize scattered knowledge like expert developers 🧠I have a tutorial about building intelligent AI memory systems with Cognee in my 'Agents Towards Production' repo that solves a critical problem - developers navigate between documentation,
1
3
7
Read the full write-up (including an example) here: https://t.co/YXtmEdU6dk
cognee.ai
Build richer AI memory with weighted nodes and edges—enable memory layers and context engineering for recency and importance. Try Cognee for enhanced reasoning.
0
0
3
I keep hearing: “Is there a way to add weights to memories and change priority?” Yes, there is. With @cognee_ , your semantic memory isn’t just connected—it’s prioritized, time-aware, and source-sensitive. We introduced advanced weights: multiple named weights on nodes/edges
2
1
9