In classic RAG, retrieval is one step in a fixed pipeline: user asks → retrieve docs → stuff into prompt → generate answer.

In agentic systems, retrieval is something fundamentally different: a core capability the agent actively decides to use (or not).
The agent chooses when to search, what to search for, which retrieval tool fits the job, and whether the results are good enough or if it needs another pass.

Join us for an evening of conversations about this new role for retrieval inside AI agents.

Agenda:

18:00: Doors open, welcome drinks
18:20: Opening keynote from Qdrant's CTO Andrey Vasnetsov.
18:55: Panel driven by your questions
19:50: Networking with refreshments
21:00: Wrap up

Panel "Retrieval in the Age of Agents"

Our panel will debate what it actually takes to make retrieval a useful (and used) tool in production agents:

Qdrant (host): the search engine as the context layer for agents;
Cognee: hybrid agentic memory architectures;
deepset (makers of Haystack): retrieval orchestration for production agents;
LlamaIndex: leading document agents and OCR platform;
n8n: low-code orchestration for agentic workflows.

After the panel, we'll move into networking with refreshments.

What We'll Debate

When and how agents should retrieve, and the right retrieval tool for the job (graphs, vectors, lexical, grep, and beyond).
The evolution from "give an LLM some chunks" PoCs to production-grade agentic systems built on real retrieval.

You drive the panel!
The most-voted questions from the registration form get answered live, so the questions that have been bugging you are the ones we'll dig into.

Vector Space Meetup: Retrieval in the Age of Agents with Qdrant, n8n, Cognee, deepset, LlamaIndex