ASUS has announced an "Agent Computer" initiative aimed at making it easier to run AI models and AI agents on your own hardware rather than in the cloud. The announcement, dated 12 November 2026 in ASUS's press release and published on its news site on 13 November, covers everything from DIY motherboards to workstations, plus a set of ASUS AI apps.
Forge PC hasn't tested any of the hardware or software mentioned. Details below come from ASUS's announcement; UK pricing and availability were not included.
What ASUS announced
According to ASUS, the Agent Computer initiative lets users deploy AI locally, in the cloud or through hybrid models, with the aim of giving them control over performance, data privacy and running costs. ASUS says it also allows some existing ASUS devices to be used as Agent Computers.
The hardware side is a grouping of ASUS products rather than one new machine:
DIY builds using current ASUS motherboards with NVIDIA GeForce RTX or AMD Radeon AI graphics cards. ASUS names AI-ready boards across its PRIME, TUF Gaming, ROG and ProArt ranges.
Workstations, including the ExpertCenter Pro ET900A X9.
Mini PCs, including the ProArt GR1X for creators.
External accelerators and enterprise systems.
ASUS has also launched a "Find Your ASUS Agent Computer" tool that recommends hardware based on workload, performance needs and whether you want local, cloud or hybrid AI.
The software
ASUS highlighted three software components:
ASUS Zenni Claw, which ASUS says makes agentic AI practical on consumer devices.
ASUS Multi-LM Tuner, to help users deploy optimised large language models across ASUS platforms.
ASUS DisplayWidget Center, which ASUS describes as a bridge between its displays and agentic AI workflows.
ASUS's announcement doesn't give technical details such as which models or runtimes the tools support, so it's too early to say how they compare with established free tools like Ollama and LM Studio.
Part of a wider trend
ASUS isn't alone. GIGABYTE has been building out its AI TOP range, including the AI TOP 100 B850 desktop, which its September press release says pairs a 16-core Ryzen 9 9950X and 128GB of DDR5 with either an RTX 5090 or two Radeon AI PRO R9700 cards for a combined 64GB of graphics memory, and the RTX Spark-based AI TOP AERO. The common thread is graphics memory: running bigger models locally needs more VRAM or unified memory.
What it means for UK builders
For now, this is more a branding and software push than new hardware for DIY builders. If you already have a recent motherboard and a graphics card with plenty of VRAM, you can run local AI models today with free tools; see our guide to running AI models locally. For buying advice, our best PC for local AI and best GPU for local LLMs guides explain what matters most.
We'll look at ASUS's software once it is more widely available and documented.
Our verdict
ASUS's Agent Computer initiative bundles its motherboards, RTX and Radeon AI builds, mini PCs and workstations under one local-AI banner, adds a hardware recommendation tool, and introduces Zenni Claw, Multi-LM Tuner and DisplayWidget Center software. It doesn't change what hardware you need for local AI, and ASUS hasn't published UK prices or detailed software specs. Forge PC hasn't tested any of it; the details come from ASUS's 12 November 2026 announcement.