You do not have to give your AI work to Deloitte, PwC, EY or KPMG. For a UK firm of 50 to 1,000 staff, the realistic alternatives are mid-tier advisory firms, specialist AI consultancies, small fixed-fee builders, freelancers and vendor professional services, and the right choice depends on the size of the programme, who your auditor is, and how much of the price goes on people you never meet.
This page is for the COO or CTO who has a Big Four proposal on the desk, or assumes the work has to go to one. It covers when the Big Four are genuinely the right call, what the other options look like, what the large firms publish as day rates, the auditor independence rules, and how to compare bids on equal terms. If you want a longer list of named firms, see our guide to the best AI consulting firms in the UK.
When a Big Four firm is the right choice
Be fair to them. The Big Four are good at things that smaller suppliers cannot match. They can put fifty people on a programme next month and keep them there for two years. They work in almost every country you might trade in. They have regulatory, tax, risk and audit specialists under one roof, which matters if your AI work touches financial reporting, a regulated product or a cross-border data question. And a board can point to a brand that investors, lenders and regulators already recognise.
So if you are running a large transformation across several business units and countries, need audit-grade assurance over an AI control, or need a supplier whose balance sheet can stand behind a multi-million pound contract, a Big Four firm is a sensible shortlist entry. The question is whether your project looks like that. Many mid-sized firms are buying something smaller: a readiness assessment, a pilot on one process, or a single production system.
The realistic alternatives
Mid-tier advisory firms such as BDO, Grant Thornton, RSM and Forvis Mazars (the name Mazars adopted when it joined with FORVIS on 1 June 2024) offer the same broad mix of audit, tax and consulting at a smaller scale. They often serve mid-sized companies as a core market, so you may get more senior attention. The same auditor independence rules apply to them as to the Big Four if they audit you.
Specialist AI consultancies do only AI and data work. They range from teams of a few people to a few hundred. They tend to be closer to the technology and less tied to a methodology, but you should check depth of bench and how they handle continuity if a key person leaves. Our comparison of AI consultancy team sizes goes into the trade-offs.
Boutique, fixed-fee builders are small teams that scope a defined piece of work and price it as a whole. You know the cost before you start, and the people who sell the work usually build it. The trade-off is capacity: they cannot staff a large programme. Freelancers are the cheapest route per day and can be excellent, but you take on the project management, quality control and continuity risk yourself. Vendor professional services (the consulting arms of software and cloud providers) know their own platform deeply, but they are not neutral about which platform you should use.
| Type | Typical engagement | How pricing usually works | Who does the work | Best for |
|---|---|---|---|---|
| Big Four (Deloitte, PwC, EY, KPMG) | Large, multi-phase, often multi-country programmes | Time and materials on a rate card, sometimes fixed price with contingency added | Mixed pyramid, partners lead, consultants and offshore teams deliver | Large transformations, regulated or audit-linked work, board cover |
| Mid-tier advisory (BDO, Grant Thornton, RSM, Forvis Mazars) | Mid-sized programmes, often bundled with other advisory work | Time and materials or capped fees | Smaller pyramid, more senior time per client | Mid-sized firms that want a recognised name and broad advisory in one place |
| Specialist AI consultancy | Strategy through to build for AI and data | Mix of fixed fee and day rate | Technical specialists, smaller teams | Firms that want depth on AI rather than general consulting |
| Boutique fixed-fee builder | Assessment, pilot or one production system | Fixed fee per phase | The founders or a small named team | Defined projects with a clear success test |
| Freelancer | A specific role or task | Day rate | One person | Firms with in-house project management that need a skill gap filled |
| Vendor professional services | Implementation of the vendor's own platform | Day rate or bundled with licences | Vendor engineers or partners | Firms already committed to one platform |
What the large firms publish as day rates
Big Four rates for private clients are rarely published, so do not trust any article that quotes them with confidence. The best public benchmark is the UK government's G-Cloud 15 framework on the Digital Marketplace, where suppliers file rate cards that public bodies can buy from. Most of the cards below were filed in January 2026; PwC's AI services card is from G-Cloud 14 in May 2024. They are public-sector framework rates, exclude VAT, and are not what a private buyer will necessarily be offered. Treat them as a reference point for the seniority you are paying for, not a price list.
| Supplier or market | Document | Lowest published day rate | Highest published day rate | AI-relevant example |
|---|---|---|---|---|
| Deloitte LLP | G-Cloud 15 Standard Rate Card | £765 | £2,125 (specialist rate up to £2,340) | Data scientist £1,275; senior machine learning engineer £1,530 |
| KPMG LLP | G-Cloud 15 Rate Card (stated as maximums) | £840 | £2,808 | Data scientist £1,080; head of data science £2,808 |
| EY | G-Cloud 15 Pricing Document (SFIA levels, 7-hour day) | £448 | £2,912 | SFIA level 5 (ensure or advise) £1,764 |
| PwC LLP | G-Cloud 14 Cloud Support Rate Card for its AI services (SFIA levels, May 2024) | £100 | £2,750 | Strategy and architecture, SFIA level 5 (ensure or advise) £2,300 |
| UK contract roles citing AI | IT Jobs Watch, 6 months to 5 November 2026 | £385 (10th percentile) | £775 (90th percentile) | Median £560 a day |
Two details in these documents are worth knowing. Deloitte's card says that when the rate card is used for fixed price deliverables, contingency may be added. And EY's card defines a working day as seven hours, while Deloitte and KPMG use eight, so the same day rate buys different hours. PwC's card says its development and delivery rates assume nearshore and offshore staff, and that UK-based staff are charged at the higher change and transformation rates where security rules apply. Its separate G-Cloud 15 card for project delivery roles, such as programme managers, sets maximum onshore rates of up to £3,655 a day. For a wider view of what UK AI work costs, see how much AI consultancy costs in the UK.
If your auditor is offering the AI work
This is the point many buyers miss. The FRC's Revised Ethical Standard 2024, in force since 15 December 2024, limits what an audit firm can sell to the companies it audits. If you are a public interest entity, such as a company with securities on a UK regulated market, a bank or an insurer, your audit firm and any member of its network may only provide services on a short permitted list, with audit committee approval. Fees for those services are capped at 70% of the average audit fee over the last three years. Most AI build work is not on that list.
If you are not a public interest entity, the rules are looser but still real. For any audit client, the standard says the firm must not design, provide or implement IT systems that are important to a significant part of the accounting or financial management system, where the auditors would rely on them. It must not take on part of the role of management. Other IT work is allowed only where safeguards, such as a separate team, can be applied. An AI tool that touches invoices, revenue or month-end reporting can fall on the wrong side of that line. Speak to your audit committee before you engage your auditor, and read our piece on conflicts of interest in AI consultancy.
How to compare proposals on equal terms
Who actually staffs it. Ask for named people, their seniority and how many days each will spend. A proposal led by a partner can be delivered mostly by junior or offshore staff, which is fine if you know it and the price reflects it.
Fixed fee or time and materials. A fixed fee moves the overrun risk to the supplier. Time and materials keeps it with you, so insist on a cap and a change process. If one bid is fixed and another is a day-rate estimate, convert both to a total with the same scope before comparing. Our guide to fixed-fee AI consultancy covers this in more detail.
IP ownership. Check who owns the code, prompts, evaluation sets, fine-tuned models and documentation at the end. Some suppliers keep reusable accelerators and license them to you. That can be reasonable, but you should know what you would lose if you stopped paying.
Exit and handover. Ask what handover looks like, what documentation you get, and whether your team can run the system without the supplier. If the AI vendor underneath shuts down or changes terms, what then? Our note on AI vendor shutdown and continuity has a checklist.
Vendor neutrality. Many consultancies hold partner status with cloud and model providers, and some earn referral fees or credits. That is normal. Ask them to disclose it, and ask why they are recommending one platform over another for your case.
Questions to ask every bidder
Use the same list for every supplier, Big Four or not. Who are the named people on this work, and what share of total days does each spend? Is the price fixed, capped or open, and what triggers a change request? Who owns every piece of the output at the end? What does a handover look like if we part ways after the pilot? Do you audit us or any company in our group, and has your independence team cleared this work? Do you earn anything from the vendors you are recommending? What does success look like at the end of phase one, and how will we measure it? For a fuller framework, see how to choose an enterprise AI consulting partner.
Where a small fixed-fee team fits
Kelriva AI is a two-person London consultancy and sits in the boutique, fixed-fee group. We publish our prices: an AI Readiness Assessment from £3,500 over 9 to 14 days, a pilot from £15,000 over 4 to 8 weeks, a production build from £40,000, and a day rate of £950 to £1,500. Our work for bettercoach produced a privacy-aware matching engine that cut the time to match a client with a coach to seconds. We are not the right choice for a programme that needs a large team at once, delivery in many countries, a 24-hour managed service or audit-linked assurance. For those, a Big Four or mid-tier firm is the better fit. If you want a quick, independent view of where you stand before you sign anything, try our free AI readiness score.
Sources
All links checked on 5 November 2026. Deloitte LLP, G-Cloud 15 Standard Rate Card, January 2026: https://assets.applytosupply.digitalmarketplace.service.gov.uk/g-cloud-15/documents/92485/484105941903512-pricing-document-2026-01-21-1142.pdf. KPMG LLP, Rate Card for G-Cloud 15, January 2026: https://assets.applytosupply.digitalmarketplace.service.gov.uk/g-cloud-15/documents/93303/656819512444092-pricing-document-2026-01-28-1150.pdf. EY, G-Cloud 15 Pricing Document, January 2026: https://assets.applytosupply.digitalmarketplace.service.gov.uk/g-cloud-15/documents/92648/505935151568640-pricing-document-2026-01-29-1021.pdf. IT Jobs Watch, AI contract job trends, 6 months to 5 November 2026: https://www.itjobswatch.co.uk/contracts/uk/artificial%20intelligence.do. Financial Reporting Council, Ethical Standard for Auditors (Revised Ethical Standard 2024, effective 15 December 2024): https://www.frc.org.uk/library/standards-codes-policy/audit-assurance-and-ethics/ethical-standard-for-auditors/. ICAS, FRC publishes Revised Ethical Standard 2024: https://www.icas.com/news-insights-events/news/ethics/frc-publishes-revised-ethical-standard-2024. PwC, G-Cloud 14 Cloud Support Services Rate Card, May 2024: https://assets.applytosupply.digitalmarketplace.service.gov.uk/g-cloud-14/documents/92454/608947631153362-pricing-document-2024-05-07-1221.pdf. PwC, G-Cloud 15 Lot 3 Ancillary Services Rate Card, January 2026: https://assets.applytosupply.digitalmarketplace.service.gov.uk/g-cloud-15/documents/92454/830153577106392-pricing-document-2026-01-30-1339.pdf. Forvis Mazars, Mazars and FORVIS to form unique global network: https://www.forvismazars.com/uk/en/who-we-are/news/mazars-and-forvis-to-form-unique-global-network.