Founder of @cognee_ the original OSS Agent memory platform. Econ + Psychologist by training. Solo-bootstrapped → backed by Pebblebed. Berlin. https://xitter.cf/t.co/1HTK5ti0T4
Berlin
Joined February 2014
- Tweets4.3K
- Following1.8K
- Followers7.2K
- Likes4.4K
Awesome event!
Last night’s @daytonaio AI Builders with @Oracle in SF was special! 🦄🚀
Our first event as a unicorn — full house, great talks & networking!
Huge thanks to Balraj Kadaikal & Srikanth Kodali (@Oracle), @TomHacohen (@SvixHQ), @tricalt (@cognee_) & @ivanda_mislav (@daytona), everyone who joined us!
See you at the next one!
Congratulations to the Daytona team!
A huge milestone
We just raised a $60M Series B at a $1B valuation to build @daytonaio into the Agent Cloud.
@FirstMarkCap led the round, with all key existing investors and angels doubling down.
We're also thrilled to welcome the founders of some generational companies aboard, including @patrickc (Stripe), @tobi (Shopify), @AravSrinivas (Perplexity), @BrendanFoody (Mercor), @oliveur (Datadog), @adamdangelo (Quora) and @ericsyuan (Zoom)
We're a 35-person team. But that's about to change; we're hiring across the board, and we'd love for you to join us.
Vasilije retweeted
RAG and agent memory get used as synonyms. They aren't.
RAG reads from a corpus that someone else wrote: docs, a wiki, a product catalogue. The agent only reads.
Agent memory starts when the agent writes back:
→ a preference the user mentioned
→ the outcome of a task it just finished
→ a correction someone made to its answer
Writing back changes the problem. You now decide what is worth keeping. You merge the new fact with what's already stored, and you handle the old fact it contradicts.
Cognee's agent memory guide puts it in one line: "The moment your system starts writing back — capturing facts, recording previous interactions, or storing the outcomes of its own work — you've moved beyond RAG into agentic memory."
Read-only retrieval is a search problem. Read-write retrieval is a data-management problem.
Read more here: cognee.ai/blog/fundamentals/…
Vasilije retweeted
@bot LDN Community Hack is this Thursday, and @cognee_ is a tech partner.
Come build with @grok + @cognee_ this Thursday.
@bot LDN Community Hack by @frankterpo, with @supabase, @zero_gtm and @cursor_ai as tech partners and @Expedite_Now hosting.
Eleven London communities are taking part.
Free, last spots open:
luma.com/eveur09a
Someone saw my upcoming @daytonaio event talk on the 20th and added Cognee to their sandbox product and sent me a pitch
Sandbox market is heating up.
Weekly update, cognee 1.6.5 is out today with 90 000 lines of code added:
Code Graph: Docstrings and execution flows are now ingested as first-class graph documents linked to the code symbols they describe, and we added search and navigation operations designed for coding agents. You can also index any branch of a repo, not just the default one. Powered by Enola
Costs became visible. Embedding tokens are now metered separately from LLM tokens across every embedding engine
Lower friction, more sources. Copying .env.template now gives you a fully keyless local setup
Google Calendar joined the connector lineup with OAuth support, right behind last week's Notion source.
The ecosystem keeps growing. The cognee plugin for Codex is now live in OpenAI's plugin directory, and our n8n node became true long-term memory for AI agents the week before.
Also in this release: a published vulnerability disclosure policy, automated daily CVE monitoring, and a batch of retrieval-accuracy and reliability fixes — 40+ tickets closed in total.
Happy that Harvey re-discovered graphs
We've been at it for 3 years
Looking forward to those benchmarks
We built context graphs to help agents work through long-horizon legal tasks.
Starting with Legal Agent Bench tasks in 11 categories like M&A diligence and contract negotiation, we had agents build context graphs of the underlying documents (without access to the eval tasks or grading rubrics).
Context graphs dramatically improved agent performance on LAB tasks:
- Firm Knowledge tasks: 65% → 80%
- Precedent lists tasks: 67% → 80%
- Matter counting tasks: 53% → 83%
They also boost agent efficiency: we found a 48% cost reduction for agents given access to a context graph to assist their work.
Full video below - more to come soon.
Looking forward to the event!
Four corners in the ring this Saturday: @Claude_Memory, @helixdb, cognee, and whatever memory setup you bring.
Same scenarios for every system, results live. Mem-Off, San Francisco, Sat 17 Oct.
Free: luma.com/cognee-cy86
Met him a week ago. Tried to quiz me on memory
🚨An employee of OpenAI has been arrested for and charged with using or endeavoring to use an electronic, mechanical, or other device to intercept oral communications.
🔗: justice.gov/opa/pr/saudi-nat…
Looking forward to the event!
Join us on Saturday, October 17 in San Francisco for Mem Off, the AI memory showdown with @Claude_Memory and @helixdb.
Everyone claims their AI has memory. Bring yours and run it on the same scenarios as everyone else's.
Free: luma.com/cognee-cy86
