Go Local.ai

Product — Audit

AI Cost Saving Audit

Our full cost saving audit can end four ways

We only profit further from two

No open ended billing, this fixed price assessment looks at what your current AI use actually costs, whether open weight models are effective for your use case, where you could save, including if you decide not to change anything about your current implementation. 
No matter the outcome there is no commitment to go further with us, you keep the report and everything produced from it.

Go Hybrid

The majority of your work moves to an open weight model, hosted either by you or by us at one of our partner data centres. Unexpected Peaks, difficult or exceptional cases are routed automatically to a cloud model. You can save costs exactly where it matters most.

Paid work follows

Go Local

Inference shifts from cloud models entirely. Either onto compute you run yourself or hosted by us at one of our partner data centres. This can be as easy as switching which endpoint your workflows connect to, just without per token fees attached.

Paid work follows

Stay where you are

Your usage may be too low, too uneven or your workloads genuinely justify frontier pricing. Moving is not always the right answer. In some cases it would cost more than it would save.

Nothing further to buy

Not yet

Your documents, data or process may need organising or reworking before they can be used efficiently in any AI workflow. It is better to discover now than partway through a migration.

Nothing further to buy.

Where the answer is hybrid or local, the next step is a paid pilot on one of your workflows. After which you can either host with us or we can work with you on setting local inference up on your own hardware. The cost of this audit is credited towards any future pilots or implementations with us.

Our reports are inspired by clinical trial methodology.

Anonymised data, blind testing, and the standards a model has to meet in writing before everything else.

Minimising bias, 
Maximising honesty.

What the report includes

Your real AI spend

Rebuilt from invoices, API costs, and token overspends. Including that subscription you forgot to cancel months ago.

Cost per unit of work

How many tokens you’re spending on each document drafted, description written, or report analysed.

Your AI workflow mapped

Team, task, model, volume, and where you’re spending what.

Where your data is going

Which data, from what teams, databases, and other sources end up in cloud models outside your organisation’s oversight. What country it sits in and whether it can be used to train on.

Open weight Model recommendations

An assessment of which open weight model or combination of models fits your workflows best, with our reasoning and recommendations for configuration.

Performance measured

Inference speeds, time to first response, token generation, and everything else that could slow down your workflows measured and compared.

Blind test results

Open weight models’ output against your current provider. On a sample of your real work, labels are removed, your team scores it.

Cost projections and Comparisons

What each outcome would cost you, over what period, and how that would change as your business grows. Including when it would make sense to switch over to a different conclusion.

An environmental impact estimate

That you can check yourself, energy usage measured where we can, on the potential hardware, grid intensity, CO₂ output, and what you can do to improve your impact.

And everything else you can cut.

Dormant seats, unnecessary tools, overinflated prompts. Your information to act on even if you don’t proceed with us.

We sell hosting, so why can you trust us to be objective during the audit?

We take every measure to maximize objectivity throughout the audit, ensuring the truth in data is presented to you. Only verifiable facts determine the outcome of the audit.

The scoring rubric is agreed to before tests are run

What results in each of our outcomes is clearly laid out and agreed to mutually. This stops us from being able to rationalise a poor result into a different outcome.

Findings will be shared

Every piece of information gathered is handed over after the audit, so either you or a different provider can verify our conclusion from the source data.

Each claim, number or statistic is traceable

Whether it’s an invoice line, dashboard metric, or logged run, we will inform you how we determined our scores and how we support our conclusions.

Every issue is communicated

Everything that goes wrong is tracked, analysed and rated on severity, recurrence risk and compared to your current workflows.

You’re the judge

When it comes to comparing current providers versus open weight models, output quality is anonymised and scored by your team. You’re the experts when it comes to your own data, we can’t fudge blind preference.

Our no migration rate is public

See our transparency page for how many previous clients we did not find enough cost savings warranting a migration. No-migration rate

What the audit does not do

We don’t change your workflows

Implementation, hosting and a pilot are separate. The audit just covers analysis, cost savings suggestions and a potential plan.

Not every workflow

Three are explored in depth in the standard fee, extras are a published unit rate.

No legal or compliance opinion

We map your data flows, find gaps, tell you where it ends up, and point out potential risks. But we are not solicitors, and cannot determine for you if it is compliant to your industry standards and regulations.

No guaranteed saving

It would be dishonest to say we will always find somewhere for you to save costs. Every case is unique, but you’re doing this audit for a reason.

No lock in

The report, data, and results are yours, you can implement them yourself, hand them over to a different provider or turn them into confetti.

Pricing. One payment, two tiers.

Lite £1,800

Your real AI costs, data, workflows mapped on paper. No Verdicts, just possible savings and usage.

Full £3,000

Everything, including the blind test on samples of your real work. Every workflow mapped, three in depth in the standard fee.

Whichever tier you pick, the price is fixed and quoted in writing before we even sign any NDAs and DPAs. No day rates, and if the work takes us longer than estimated we absorb the costs.

Lite is credited against the price of a full audit.

Your invoice does not change over the course of the audit.

Lite

1 week

Full

2 weeks

Find out more

Give us a quick call, no charge. We can work out if you might be overspending on AI or might be better served by local inference. Or if you're curious about any of our promises and claims we'd love to explain those further.

Email us

Send us the details of what you're working with and what you're hoping to change. We reply to every enquiry.

contact@go-local.ai

Book a meeting

Pick a time that suits you for a free 30-minute consultation. No commitment, no sales pressure, just a conversation about where you are.

Book a 30-minute call