Foundry
Coming soon. LLM Foundry is being prepared for early Fit Reviews. To register interest, email info@llm-foundry.com.

Offers

Advice. Managed setup. On-prem support.

Start with a practical fit review, move to a managed private AI process, or choose deeper support for teams that need stronger control and records.

Book a Foundry Fit Review

Foundry Advisory

£299 one-time

First, find out whether this is worth doing.

Best for: Owners, partners, MDs, operations leads, and CTOs who want a clear answer before buying hardware or starting a project.

  • A review of one workflow
  • What data is involved
  • What needs to stay controlled
  • Where approvals are needed
  • Whether Foundry is likely to save time or money
  • What hardware may be needed
  • Whether the answer should be yes, no, or start smaller

Outcome: A clear recommendation before you buy hardware or start a project.

Book a Foundry Fit Review

Foundry Managed

£999 setup + £99/month support

A managed private AI workflow for one real business process.

Best for: Businesses ready to move one suitable job into a private setup.

  • Setup on suitable Apple hardware
  • A workflow built around one agreed job
  • The right tools for each step
  • Review queues and approval points
  • A practical record of what happened
  • Testing with sample work
  • Handover and ongoing support

Outcome: A private AI process your team can use, check, and improve.

Hardware is not included. We advise on the right Apple specification if you do not already own suitable hardware.

Discuss Managed setup

Foundry On-Prem

£2,000–£5,000/month

For teams that need stronger control, records, and support.

Best for: Legal, finance, healthcare, government-adjacent, defence-adjacent, or enterprise teams with stricter data, records, confidentiality, procurement, or continuity requirements.

  • Local deployment behind your business boundary
  • Clear records of key workflow steps
  • Approval gates for sensitive outputs
  • A runbook for how the process works
  • Support boundaries and escalation route
  • Regular review of the workflow and tools

Outcome: A private AI setup that managers, IT, operations, and risk/compliance stakeholders can understand.

On-Prem support can help with evidence, documentation, and control design. It does not certify regulatory compliance or replace legal, data-protection, security, or professional advice.

Discuss On-Prem requirements

Clear boundaries prevent expensive surprises.

  • Hardware purchase: you buy the Mac Studio or approved Apple hardware.
  • Replacement of existing business systems: email, helpdesk, CRM, case management, accounting, GitHub, and backups stay in place.
  • Open-ended custom development: new jobs are scoped separately.
  • Unsupervised legal, financial, medical, HR, compliance, or customer-facing decision-making.
  • Training your team to become private AI engineers.
  • Claims that every workload should move into Foundry.
  • Legal advice, data-protection advice, SRA advice, DPIA completion, or compliance certification.
  • Guarantees that every future model, licence, or hardware generation will be supported.

FAQ.

Is Foundry just ChatGPT in the office?

No. Foundry is a managed private AI process for repeat work. It keeps sources, records, review steps, and approvals together.

Does everything stay local?

For agreed workflows, the AI work can run on equipment your business controls. Your normal email, case system, helpdesk, backups, or other tools may still use their existing cloud services.

Does this make us GDPR or SRA compliant?

No product can promise that. Foundry can help you keep better records, approvals, and control around AI work. You still need your own legal, compliance, contracts, policies, and professional judgement.

Does Foundry use one AI model for everything?

No. It can use the right tool for each part of the job: fast tools for simple sorting, document tools for extraction, and stronger AI for careful drafts or summaries.

Can it improve later?

Yes, where the hardware, licence, workflow, and support position allow it. The process can be improved as better local AI tools become practical.

Is it suitable for law firms?

It can be suitable for confidentiality-sensitive work such as intake, document review, knowledge search, and drafting support. For legal and professional-services work, Foundry can support better records, source checking, and approval steps. It does not make a workflow automatically compliant, confidential, or privilege-safe. Your own obligations and advice still matter.

Can we use our existing Mac Studio?

Possibly. The Fit Review checks the hardware, memory, storage, workload, and expected capacity requirements.

Do you support Windows or Linux?

Not for the MVP offer. Foundry is positioned around Apple hardware.

Will it replace staff?

No. Foundry is designed to reduce repetitive first-pass work and queue drafts for review. It does not replace judgement, client care, legal advice, engineering ownership, or operational accountability.

Can it run multiple workflows?

Yes, if the hardware and support plan can handle them. The Fit Review helps define what fits safely.

What if the workflow is not suitable?

We will say so. Sometimes the right answer is: keep it manual, keep it cloud, clean the data first, tighten the approval process, or start with a smaller job.