Is on-prem right for you? A two-minute self-check

Answer six questions and the self-check suggests a small, medium or large starting tier for running AI on your own premises.

Six questions, about two minutes. Your answers stay in this browser tab: nothing is sent, stored or tracked.

1. How sensitive is the data you would want AI to work with? Mostly public or low-risk material Internal commercial data (contracts, pricing, plans) Highly confidential IP, personal data or client-privileged material

2. How often would your team use AI for routine questions and drafting? Occasionally Most days, across several people Constantly, or as part of automated workflows

3. How many people would use it? A handful One or more teams Much of the organisation

4. Do regulators, clients or contracts set rules on where your data may go? No specific rules beyond general good practice General data protection law applies (for example HK PDPO or UK GDPR) Sector rules or client contracts also restrict data location or third-party processing

5. How would you prefer to pay for AI capacity? Pay as you go, with little or no upfront spend A one-off workstation-class purchase is acceptable We can plan for server-class equipment with ongoing support

6. Who would look after the equipment day to day? Nobody in-house at present A small IT function or an external provider A dedicated IT or infrastructure team

Disclaimer. This self-check is a simple, illustrative guide to start a conversation. It is not advice and does not recommend any product, vendor or configuration. The scoring bands are our own rule of thumb, not an industry standard. The right answer for your business depends on factors a short questionnaire cannot capture, so each business needs its own assessment.

The three starting tiers

Small: one workstation-class machine running a compact open-weight model for a few users, or staying with hosted services for now. A low-cost way to test value.
Medium: a high-memory workstation or small server shared by one or more teams, with access control, logging and backups set up from the start.
Large: server-class equipment with ongoing support, sized for larger models or many users, and governed as part of your wider IT estate.

Each tier is described in more detail, with example models and sizes, in AI hardware for on-premises use. For the security side, see security and compliance for on-premises AI.

Talk to ISD

If you would like a second opinion on your result, we are happy to talk it through, with no obligation. Email [email protected] or use the contact details on the ISDept.ai home page.

← Back to ISD Insights

Is on-prem right for you? A two-minute self-check

The three starting tiers

Talk to ISD

Questions about what this means for you?