
What happened
FrigadeHQ's Hob installs on Windows, Mac and Linux, scans the PC, and offers an AI model suited to its memory and GPU; the reviewed Windows 11 build ran CSV-to-chart requests, PDF summarization and folder projects, and adds Qwen3-VL for images.
Why it matters
By removing model selection and runtime setup, Hob lowers the barrier to running AI locally on an ordinary PC, and its offline operation keeps data on the device.
Summaries like this, every morning.
Hob is aimed squarely at the setup work that keeps many people from running AI on their own hardware. Rather than asking users to pick among model sizes and build a runtime, it inspects the machine and downloads a matching open-weight model such as Gemma or gpt-oss, then opens a chat window. On the reviewed Windows 11 machine, that meant attaching a sales CSV and asking for a monthly bar chart with a Japanese explanation, and attaching a PDF with a request for three key points — both of which the app handled. Image analysis is delegated to a separately downloaded Qwen3-VL model.
The app also organizes work beyond single chats: a folder can be registered as a project so its files are handled together, and chat history can be saved as Markdown. It connects to outside services through MCP, a common standard for letting AI call external apps, though such links may require authentication and network access.
Reaction on Hacker News was mixed. One commenter welcomed having something that simply works, while others questioned how capable the model is and noted the lack of a straightforward way to remove the 18GB model. Hob deliberately stops short of deep tuning: it supports changing inference depth and the model in use, but not Temperature, context length or GPU usage. That places it for people who want quick local chat and file handling rather than fine-grained control.
Pick your industry and the AI tools you use, and get news related to your work every day.
Free · 30 seconds with Google · unsubscribe anytimeWhat is AIToday? →
Ask AI anything about this article. The AI reads this article, earlier AIToday articles, and Wikipedia, and cites its sources. Q&As are published on this page for other readers too.
Atomicwork launched an AI Coworker Marketplace with certified AI coworkers for cloud infrastructure, access ma…
Amazon is adding a texting feature to Alexa for Shopping, letting US customers describe or photograph items an…

Carnot AI Inc, which develops and provides the enterprise AI platform Jinba, was accepted into NVIDIA Inceptio…

In a guest commentary, R K Anand of Tensordyne argued challengers should target inference—not training—citing…

Thinking Machines released Inkling 2 Small, a 594B-total / 17B-active multimodal MoE, available now on Tinker

Anthropic launched Personal Finance as a beta in the US for Pro, Max, and Ultra subscribers, giving Claude rea…
