AI

UiPath is expanding its Automation Suite with agentic AI capabilities, enabling government institutions and regulated industries to deploy autonomous systems directly within their own secure environments — without sending data to the cloud.

A new McKinsey report shows that agentic AI can boost productivity in insurance system modernization by up to 90%. Why are Czech insurers still hesitating and where are the biggest savings?

According to reports from analyst Ming-Chi Kuo, OpenAI is accelerating development of its own phone with the goal of integrating AI directly into the hardware. What does this mean for the future of mobile carriers and users in the Czech Republic?

Hermes Agent by Nous Research is an open-source autonomous AI agent with a closed learning loop that builds skills from its own experience, supports 15 communication platforms, and runs on seven different backends from local PC to cloud.

Meta is building a highly personalized agentic AI assistant powered by its Muse Spark model, aiming to take on OpenClaw from OpenAI and become part of users' daily routines. When will it launch, and what does it mean for Czech users?

Google is upgrading Gemini API File Search with multimodal capabilities and precise page citations. The new Gemini Embedding 2 model can now search documents and images simultaneously — and point to the exact page for every answer.

xAI releases Grok 4.3, its fastest and most intelligent model to date. With a million-token context window, top-tier agentic capabilities, and a price of $1.25 per million tokens, it targets developers and enterprises directly. In agentic tasks, it surpasses both Gemini and GPT-5.4 mini, while dominating the ValsAI leaderboards in legal and financial domains.

ChatGPT's new default model promises 52.5% fewer hallucinations in medicine, law, and finance. It delivers sharper answers, smarter personalization, and a new visible memory feature that puts you in control of your data.

UiPath Automation Suite now brings on-premises agentic AI to government institutions. What does it offer for Czech authorities, how does it work, and why is full data control so crucial?

New speculative decoding technique lets Gemma 4 models generate text up to three times faster without any loss of accuracy. Thanks to the Apache 2.0 open-source license, developers can try it out instantly in Ollama, vLLM, or directly on mobile.