LIVE

Testing Local AI on the Surface Laptop Ultra: What Went Wrong

💻 Laptops & PCs 1 min read 👁 2

A hands-on test of local AI agents on Windows hardware reveals major setup hurdles, unexpected automation quirks, and frustrating configuration limits.

Testing Local AI on the Surface Laptop Ultra: What Went Wrong💻Laptops & PCs

Photo: PCWorld

Key points

  • Tested local AI agents including Nous Research Hermes on Windows hardware.
  • Encountered extreme friction with permission prompts and unexpected full automation.
  • Hit roadblocks with expired API credits and failing plugin installations.

When looking for productivity gains, turning to local AI models on Windows hardware like the Surface Laptop Ultra can seem like a logical alternative to cloud solutions. Equipped with 32GB and 64GB RAM configurations respectively across test machines, these setups ran into immediate roadblocks when handling tasks using the Nous Research Hermes agent.

During initial testing, default settings proved overly restrictive, requiring manual approval for nearly every mouse movement and click, effectively negating any productivity boost. Subsequent tests on a higher-spec machine with 64GB of RAM shifted drastically to unprompted autonomous behavior due to hidden settings, while a cloud-hybrid configuration stalled entirely once API credits expired. Eventually, downloading a local 27 billion parameter Qwen 3.8 model allowed testing to resume, though basic plugin installations still triggered persistent error loops.

Should you buy it? For users seeking seamless out-of-the-box local AI productivity tools, current software iterations still require significant patience and troubleshooting.

Source PCWorld — pcworld.com/article/3280321/local-ai-productivity-is-it-more-troubl… ↗

Contact the newsroom

Got a tip, a proposal or a partnership question? Write to us and we will reply shortly.