|
LAUNCHING KELAS BELAJAR PYTHON AGENTIC AI di @WPUCOURSE
WPU
WPU-kanalen lancerer et nyt onlinekursus kaldet "Kelas Belajar Python Agentic AI" på platformen WPU Course. Kurset er beregnet til begyndere og starter med Python-programmering fra bunden, med det endelige mål at bygge en AI-agent/workflow som praksiscase. Python er valgt som programmeringssprog fordi det er tæt forbundet med AI og maskinlæring, og kurset afsluttes med et digitalt blockchain-certifikat. Under livestreamen annonceres også mulighed for stipendier samt en tidsbegrænset rabatkupon i forbindelse med lanceringen.
|
|
cognee 1.0: Self-improving memory for agents
cognee
Cognee 1.0 is presented as an open-source memory platform for AI agents, claiming to handle 100-billion-token context windows with over 79% accuracy at 7x lower cost than leading models. Unlike standard vector-based retrieval used by Claude or ChatGPT, Cognee uses a dynamic graph memory system that understands relationships between data points, updates context continuously, and can resolve conflicting information autonomously without user input. A demo scenario illustrates how Cognee outperforms a standard Claude session by correctly tracking a prospect's switch from Windows to macOS and leveraging past client playbooks to inform new work. The platform was reportedly built by neuroscience researchers from Brown and Berkeley universities and is already used by over 100 companies.
|