Články z oblasti vzdělávání a poradenství

Zencoder launches Zenflow: more agents and tests, less chaos in AI coding

Zenflow aims to replace ad-hoc “chat-to-code” with an assembly line: plan, implement, test, review.…

Agentic AI fails on complex tasks: Patronus bets on “live” simulators

Patronus AI argues agents struggle with real work: interruptions, context shifts, and layered…

Zoom claims a record on a tough AI exam; critics call it an ensemble of others’ models

Zoom reported a 48.1% score on Humanity’s Last Exam. Critics argue it reflects orchestration of…

JPMorgan scaled internal AI without mandates: connectivity to systems made it work

JPMorgan says its internal assistants spread organically: hundreds of thousands of users followed…

Nvidia opens Nemotron 3: models and data as a hedge against cloud dependence

Nvidia released open Nemotron 3 models and included training data and tuning tools. For enterprises,…

Agentic AI hits architecture: design patterns aim to stop expensive pilots

The piece argues agentic AI fails more on system design than on model choice: routing, guardrails,…

AI is both weapon and shield—governance decides which

Deloitte outlines the paradox: AI accelerates attacks (deepfakes, social engineering) but also…

Agents fail when they just assist broken processes

Deloitte warns agentic AI often hits reality: companies automate poorly designed workflows. Success…

Agentic AI could cut drug development time 20–30% - with process redesign

McKinsey says agentic AI could reduce time and costs in drug development by tens of percent. The…

Listing 28 to 36 out of 88