SLMs & Agent Memory with Cognee & Fastino Labs

​Agent memory quality depends on extraction quality. If extraction is slow, expensive, or imprecise, everything downstream inherits those problems: retrieval, reasoning, recall.

​Join Fastino Labs and Cognee for an evening on how small, specialized open-weight models can be used in the extraction layer for agent memory.

​Agenda:

• ​6:00pm - Arrival, networking, food & drinks

• ​7:00pm - Small, Fast, Local, Open: Entity & Relation Extraction for Knowledge Graphs with GLiNER2 (Mary Newhauser, Fastino Labs)

• ​7:20pm - AI Labs Are Not Your Friend, Memory & Local Models Are (Vasilije Markovic, Cognee)

• ​7:40pm - Live demo: GLiNER2 + Cognee

• ​8:00pm - Q&A, drinks, networking

​Food and drinks will be provided. 😊

​About Fastino Labs: Fastino Labs is a frontier lab headquartered in Palo Alto, California that develops SOTA SLMs (GLiNER2), fine-tuning tools, and offers optimized inference.

​About Cognee: Cognee is an open-source AI memory engine that transforms unstructured data into persistent, queryable memory. It enables AI agents to store, retrieve, and reason over information across time and infrastructure, making it possible to build reliable, adaptive, and stateful AI systems that operate seamlessly across cloud and local environments.