AI ConsultancySenior engineers / Practical advice
Self-hosted and private AI inference specialists.
Choose the right model. Buy hardware with confidence. Understand what is holding your inference back.
We help businesses plan and improve private AI, from individual GPU workstations to small and mid-scale server deployments.
Engineers who build and run AI products
We build Krasis, Lore and Dokima. Our work spans inference engines, agent workflows and AI-powered code security. That gives us practical experience of the choices around running useful AI.
Services & pricing
Start with the decision you need to make. Each assessment has an agreed scope, a defined deliverable and time with a senior engineer.
What should we run, and what should we buy?
A practical plan for a new or changing deployment. We review your workload, model options, budget and infrastructure requirements before you commit to hardware.
Can this model do the work we need?
Test a particular model against the tasks that matter to your business. We agree the examples and assessment criteria with you, then run a bounded evaluation of its capabilities and limitations.
What is limiting our inference performance?
Understand how a deployed model behaves under your workload. We assess the baseline, investigate bottlenecks and carry out agreed optimisation experiments where useful.
Does this design stand up to scrutiny?
An independent second opinion before you commit. Bring an existing architecture, proposed deployment or supplier specification and we will examine its assumptions, constraints and trade-offs.
05 / Ongoing advice
A senior engineering sounding board for decisions about your private AI systems.
Discuss an adviser arrangement£950/month
Get an experienced second opinion before committing time and budget. Where deeper investigation would help, we’ll recommend and scope the appropriate assessment.
Our consultancy services are available to business customers acting in the course of their business. All engagements are subject to an agreed scope and our Consultancy and Advisory Terms and Conditions.
All prices exclude VAT where applicable. Assessment scope and project fees are agreed before work begins, with 50% payable upfront and 50% on delivery of the agreed assessment. Any additional compute costs are agreed in advance. The adviser is a monthly service.
We commit to the agreed assessments, advice and deliverables. Specific performance improvements are not guaranteed. Implementation and ongoing operation require a separate scope.
Monthly adviser allowances do not roll over. Technical reviews cover a specific question and reasonable follow-up clarification. Benchmarking, detailed investigation and implementation are quoted separately. The adviser service does not include incident response or ongoing system operation.
How it works
You can get useful advice without giving us access to your servers. Your team can supply the configuration and run agreed measurements while we interpret the evidence.
Tell us what you want to run, your constraints and what you need to understand. We identify the assessment that fits.
We define the workloads, evidence, calls and written deliverables, then agree the price before starting.
We present the findings, explain the trade-offs and give your team recommendations they can act on.
Start a conversation
Tell us about the workload, the decision you face and any models or hardware you are considering. If you are unsure which service fits, start with the problem.
Email us about your projectThe button opens your email application. You can also copy the address into your preferred email service.
