Hardware Guide

We handle the hardware. You choose how much power you need.

Every Worker AI Pro system runs on hardware we supply, configure and install on-site. Your data stays in your building. We handle everything from day one.

ISO 9001 Certified
Cyber Essentials Certified
Capital Web Ltd, est. 2015
On-site across Kent & London
Choose Your Setup

Which setup is right for your team?

We match the hardware to your team size, workload and data privacy needs. Here is how it typically maps out.

Small team

Mac Mini M4

1 to 10 staff

Compact, silent, sits on any desk. Handles 1 to 2 AI agents and all everyday automation tasks. Perfect for businesses getting started with AI automation.

From £599 est.
Hardware only. Installation quoted separately.
Enterprise

Mac Studio Ultra or Linux Server

Enterprise or high-privacy

Full offline capability. Zero cloud dependency. For regulated industries, large organisations or businesses that need frontier-class AI with no data leaving the building.

Quoted on scope
Hardware, installation and configuration quoted together.
Mac Mini M4 -- RAM configurations
RAM Chip Est. price Model size range Best for
16 GB Base M4 £599 est. 3B to 7B models Basic chat, Q&A, single-agent automations. Suitable for experimentation.
24 GB Base M4 £799 est. 7B to 14B models Most practical business models. Good for 1 to 2 agents on HR, marketing and operations.
48 GB M4 Pro £1,499 est. 32B models + large context Longer conversations and larger documents. Consistent output quality for teams needing reliability.
64 GB Top spec M4 Pro £1,799 est. 32B sweet spot + 70B possible Best Mac Mini for business. Run one 70B model plus supporting models (routing, embeddings).
Mac Studio -- RAM configurations
RAM Chip Est. price Model size range Best for
36 GB M4 Max £2,099 est. Up to 32B models Entry Mac Studio. Good step up from Mac Mini. 2 to 3 agents simultaneously.
128 GB Great M4 Max ~£3,799 est. 70B+ models Heavy multi-agent deployments. Consistent high-quality output at scale.
96 GB M3 Ultra £4,199 est. 70B to 100B models Entry Ultra. Nearly 2x the cores vs M4 Max. Handles 70B models with ease. Multiple concurrent agents.
256 GB M3 Ultra ~£6,499 est. 200B+ models Large organisations handling complex tasks with zero cloud exposure.
512 GB Max config M3 Ultra ~£7,999 est. 600B+ models Apple's most powerful desktop. Replaces frontier cloud APIs entirely. No internet required, no per-query cost, ever.
Model sizes assume quantized open-source models (Q4 to Q5). Actual performance depends on context length and model architecture. Prices are Apple UK retail estimates at time of writing and are subject to change.
Why Local Hardware

Why local hardware over cloud APIs?

For businesses with regular AI workloads, owning the hardware is almost always cheaper over a 3 to 5 year horizon than paying cloud API fees month after month.

The ROI case

A business spending £3,000 per month on cloud AI APIs will spend £36,000 per year. A Mac Studio costs £2,099 to £7,999 once. The hardware typically pays for itself in under 6 months for businesses with regular AI workloads. From that point forward, every query costs nothing.

What you gain

  • No per-query cost once the hardware is paid for
  • Data stays in your building, not a cloud provider's servers
  • No internet dependency for local tasks
  • Consistent performance regardless of provider outages
  • GDPR-friendly by design
  • NDA available on request for sensitive deployments
Apple hardware pricing is estimated and subject to change. Prices shown reflect Apple UK retail at time of writing and may vary by configuration, availability and any updates Apple makes to its product line. Exact specification should be confirmed at point of order. Additional hardware may be required for stable operation, including UPS (uninterruptible power supply), external backup drives and network switching equipment. These are quoted separately.