Pricing
Three ways to engage, and one of them is designed to end
Start with a fixed-fee audit that is credited against whatever follows. Take implementation in phases you approve one at a time. Add a retainer only if the measurement is worth maintaining.
Fixed-fee audit
Two weeks, one price, a ranked savings plan at the end. The only sensible first step, and the one that makes every later estimate real instead of indicative.
- Priced the same whether your spend is £3k or £300k a month
- Credited in full against implementation work started within 60 days
- No obligation to continue, and the golden sets and tracing stay yours
Fixed-scope implementation
The savings plan, delivered in phases you approve one at a time. Each phase is scoped, priced and shipped against a stated saving and a stated quality guardrail.
- Phased so you can stop after any phase and keep everything built so far
- Free-saving work sequenced first, so early phases tend to fund the later ones
- Everything lands in your repositories and your cloud account
AI FinOps retainer
Monthly stewardship once the first round is done: the measurement stays honest, the guardrails stay green, and the savings backlog keeps moving.
- Named engineer, monthly unit economics pack, standing savings backlog
- Three-month minimum, then rolling with 30 days' notice
- Gain-share option on larger estates, measured on cost per successful task
By engagement
Indicative prices per service
Implementation is scoped from the audit, so these are starting points rather than quotes. Every one of them is a fixed price agreed before the phase begins — we do not run open time and materials on cost work, because that would be an odd incentive to hold.
| Engagement | Duration | From |
|---|---|---|
| AI Spend Audit Teams spending £3k+/month on LLM APIs who can see the total on the invoice but cannot explain it. | 2 weeks | £3,500fixed fee |
| LLM Gateway Implementation Companies where more than one team, product or repository calls a model provider directly. | 3–5 weeks | From £11,000fixed scope |
| Model Routing & Evaluation Products where one flagship model currently answers everything, from classification to long-form drafting. | 4–6 weeks | From £9,500fixed scope |
| Prompt, Context & Caching Engineering Chat products, RAG applications and agent loops where the same context is resent on every turn. | 2–4 weeks | From £6,500fixed scope |
| Azure AI Cost Optimisation Organisations running Azure OpenAI or Azure AI Foundry, especially where compliance rules out other providers. | 3–4 weeks | From £8,500fixed scope |
| AI FinOps Retainer Teams who have already done the first round of optimisation and want it to stick. | Rolling, 3-month minimum | From £2,400per month |
Prices exclude VAT. All engagements are UK-contracted and billed in GBP.
The arithmetic
When this is worth doing, and when it is not
Worth doing
You are spending £3,000 a month or more on model APIs, the bill is growing faster than usage, more than one team calls a provider directly, or nobody can currently say which feature is responsible for what share of the invoice. At £10,000 a month, a fifteen per cent reduction pays for the audit in under three months.
Not worth doing yet
Your spend is under a few thousand a month, you are still finding product-market fit and the workload changes weekly, or you have already been through a round of optimisation and have the measurement in place. In the first two cases the effort is better spent on the product. In the third, read the guides and see whether anything is missing.
FAQ
Pricing questions
Why is the audit a fixed fee rather than a percentage of savings?
Because a percentage-of-savings fee on an audit creates an incentive to find large savings rather than true ones, and the first thing a good audit sometimes concludes is that there is not much to save. A fixed fee keeps the finding honest. Gain-share becomes appropriate later, on implementation work, once there is an agreed baseline to measure against.
Do you offer a share of savings instead of a fee?
For estates above roughly £25,000 a month in model spend, yes: a reduced fee plus a share of verified savings over an agreed baseline. It is measured on cost per successful task rather than the raw invoice, so the incentive stays pointed at the customer outcome instead of at cutting quality. We will not do it without a jointly agreed measurement method set out before the work starts.
Is the audit fee really credited against implementation?
Yes, in full, against implementation work started within 60 days. It is not a discount that appears at the end — it comes off the first implementation invoice. If you do not continue, you keep the report, the tracing and the golden datasets regardless.
What if our spend is under £3,000 a month?
Then an engagement probably does not pay for itself and we will say so on the first call. The guides on this site cover the same levers in the same order, and most of the early wins are things a competent engineering team can ship in a fortnight without help.
Are there ongoing licence or platform fees?
None from us. Everything we build is open-source or your own cloud infrastructure, running in your subscription and billed by your cloud provider directly. There is no Rise10x software licence, no per-call fee and no hosted service in your request path.
How do you handle scope changes?
Implementation is phased and each phase is priced and approved separately, so a change of direction affects the next phase rather than renegotiating a large contract mid-flight. If work drifts outside cost engineering into general platform work, we say so and scope it separately rather than quietly billing it under the retainer.
The first call is free and frequently ends in “not yet”.
Tell us your monthly model spend and what the product does. If an engagement would not pay for itself, we will say so and point you at the guide that covers it.