Client proof

"We asked Breakthrough AI Works to audit our customer support pipeline. Within six weeks they had a classification model running that cut average ticket response time from fourteen minutes to under three. The bill was exactly what the scoping document said it would be." — Rona Gillespie, operations director, Lochside Logistics
Data-driven logistics office with AI dashboards on screens

Honest Artificial Intelligence pricing, explained line by line

Most AI consultancies sell you a package. Bronze, Silver, Gold — pick a tier and hope it covers what you actually need. We think that model is broken. It bundles work you do not need with work you do, then charges you for the bundle.

We price every engagement by the hour and by the deliverable. You see the rate, you see the scope, you see the total before we write a single line of code. If the scope changes mid-project, we re-quote the delta and you approve or decline. No surprises.

This page walks you through exactly how our costs are built, what each type of work involves, and where the money goes. Read the capability map below, then the cost breakdown, and you will know more about AI pricing than most procurement teams.

Why no packages? Packages assume every business has the same problem. A retailer automating inventory forecasts needs different compute, different data engineering, and a completely different validation cycle than a law firm classifying contracts. Bundling those into the same tier wastes your money or starves your project. We quote what you need — nothing more.

Capability map

Every row is a distinct type of work we deliver. The rightmost column shows typical engagement length — not a promise, but a realistic median from completed projects.

Capability What it covers Typical output Median duration
Data audit We review your existing data stores, schemas, quality metrics, and access patterns. We identify gaps that would block model training. Audit report with prioritised remediation steps 2–3 weeks
Model selection Given your use case, we benchmark candidate architectures — transformer, gradient-boosted tree, convolutional net, or simpler regression — on a sample of your data. Benchmark comparison document and recommended architecture 3–5 weeks
Custom model training We build, train, validate, and tune a model to your specifications. Includes feature engineering, hyperparameter search, and cross-validation. Trained model artefact, evaluation metrics, training notebook 6–12 weeks
Integration engineering We deploy the model into your production stack — API endpoint, batch pipeline, or edge device — with monitoring and rollback capability. Deployed service, CI/CD pipeline, monitoring dashboard 4–8 weeks
Ongoing model support Drift detection, retraining schedules, performance reporting, and incident response. Billed monthly at a fixed retainer you can cancel any month. Monthly performance report, retraining runs as needed Rolling monthly

How we think about your problem

Whiteboard strategy session with flowcharts and sticky notes

Before we quote anything, we spend time understanding your business outcome. Not the technical wish-list — the business outcome. "Reduce churn by 15 per cent" is a business outcome. "Build us a neural network" is a technical wish-list. The distinction matters because the cheapest path to a 15 per cent churn reduction might be a logistic regression that costs a fifth of the neural network budget.

We have turned down projects where the right answer was not AI at all. One Edinburgh retailer came to us wanting a recommendation engine; after a week of analysis we showed them that fixing their search bar ranking algorithm — a rule-based change, not machine learning — would lift conversion more. They paid us for the analysis week and saved tens of thousands on a model they did not need.

That is the kind of honesty we bring. It costs us revenue in the short term. It earns us referrals in the long term.

£110Median hourly rate across all capabilities
23Projects completed since 2022
100%On-budget delivery record
4Engagements declined on ethical grounds

Where your money goes

Below are the cost components that make up every invoice. We do not mark up cloud compute or third-party licences — you pay what the vendor charges, and we show you the receipt.

Consulting hours

Scoping, stakeholder interviews, architecture design, and project management. Billed at £95–£130 per hour depending on the seniority of the consultant assigned. You approve the staffing plan before work begins.

Typical share of total cost: 20–30%

Engineering hours

Data pipeline construction, feature engineering, model coding, testing, and deployment scripting. Billed at £100–£140 per hour. Our engineers pair-programme on complex tasks, so you may see two names on a time sheet for the same block — both are working, not duplicating.

Typical share of total cost: 40–55%

Compute and infrastructure

GPU hours for training, cloud storage for datasets, API hosting for inference endpoints. We pass these through at cost. For most mid-size projects the compute bill lands between £800 and £4,000 total. We provide a cost estimate before spinning up any resources.

Typical share of total cost: 10–20%

Third-party licences

If your project needs a licensed dataset, a proprietary embedding model, or a specific monitoring tool, we list the licence cost separately. You own the licence. We never resell software at a margin.

Typical share of total cost: 0–10%

Fit check — is this the right engagement for you?

Good fit

You have at least six months of structured data, a clear business metric you want to move, and internal technical staff who can maintain the system after handover. Budget starts around £12,000 for a scoped engagement.

Possible fit

You have data but it is messy, or you are not sure which metric matters most. We can run a paid discovery sprint (two weeks, fixed fee) to figure that out before committing to a larger project.

Not yet a fit

You are pre-revenue, have no historical data, or need a chatbot built from an off-the-shelf API with no custom training. We are happy to point you to cheaper alternatives that will serve you better at this stage.

What happens after you reach out

Three clear steps. No obligation until you sign a scope document at the end of step two.

1. Discovery call

A 30-minute video call where we listen. You describe the problem, the data you have, and the outcome you want. We ask questions. No slides, no sales pitch. If we think AI is not the answer, we say so on this call.

2. Scoping document

Within five business days we send a detailed scope: deliverables, timeline, team composition, hourly rates, compute estimate, and total cost range. You review it, ask questions, request changes. We revise until you are comfortable or walk away — no charge for the scoping work.

3. Signed engagement

Once you sign, we start. Weekly progress updates, shared task board, and a single point of contact on our side. Invoices are monthly, itemised to the hour.

Start a conversation

Fill in the short form or contact us directly. We respond within one business day.

Thank you. We will be in touch within one business day.

Prefer to talk?

+44 56 9621 9610

[email protected]

46 Jordon Rise, Nether Bergnaum Bridge, Scotland, EM4 8BT, United Kingdom

Privacy policy

Breakthrough AI Works collects personal data only when you voluntarily submit it through our inquiry form: your name, email address, chosen service interest, and message content. We use this data solely to respond to your inquiry and, if you become a client, to manage the engagement. We do not sell, rent, or share your personal data with third parties except where required by law or where a processor (such as our email hosting provider) needs access to deliver the service.

We store inquiry data on encrypted servers within the United Kingdom. Data is retained for 24 months after last contact, then deleted. You may request access to, correction of, or deletion of your data at any time by emailing [email protected]. We use a single localStorage flag to remember your cookie consent preference; no tracking cookies, advertising pixels, or analytics scripts are loaded on this site. This policy was last reviewed on 2 January 2026.

Terms of service

By using this website you agree to these terms. The content on breakthroughaiworks.click is provided for general informational purposes. It does not constitute professional advice. Engagement-specific terms, deliverables, timelines, and fees are governed by the individual scope document signed between you and Breakthrough AI Works for each project.

We make reasonable efforts to keep information on this site accurate, but we do not warrant completeness or timeliness. Breakthrough AI Works is not liable for any loss arising from reliance on content published here. All intellectual property on this site — text, graphics, layout — belongs to Breakthrough AI Works unless otherwise stated. You may not reproduce it without written permission. These terms are governed by the laws of Scotland.

Disclaimer

Pricing figures, project durations, and performance metrics mentioned on this site are indicative and based on past engagements. They are not guarantees. Every AI project depends on data quality, scope complexity, and client responsiveness, so actual costs and timelines will vary. The client proof quotation attributed to Rona Gillespie is used with permission and reflects that specific engagement; your results may differ. Breakthrough AI Works is not responsible for decisions made based on information on this website without a formal signed engagement.

This site uses a single localStorage flag to remember your consent choice. We load no tracking cookies or analytics scripts. Read our privacy policy.