Blog

How to compare AI consultancy quotes in the UK

Reviewed 25 October 2026.

Phil Patterson
calender
Published version
November 9, 2026

AI consultancy quotes are only comparable when they cover the same work. An hourly rate does not tell you who will deliver, which systems are included, how the result will be tested or who maintains it afterwards. This guide gives you a way to compare proposals without relying on unverified industry price ranges.

Separate the stages before comparing prices

Discovery or an audit clarifies the process, data, risks and opportunities. Implementation builds and tests an agreed workflow. Training prepares the people who use it. Managed support covers specified operating responsibilities after release. A proposal may include several stages, but each needs a clear boundary.

A low initial price can exclude integration, software usage, migration, documentation or handover. Equally, a larger discovery exercise may be justified where the systems or consequences are complex. Judge the stated work and evidence rather than the size of the consultancy.

Blue Canvas prices: use the current rate card

The current Blue Canvas prices page lists a fixed £750 AI audit. Builds and training are quoted after agreeing a written scope. Managed AI starts at £250 a month with a one-off setup fee from £1,750. Prices are in pounds sterling; the written quote confirms scope, timing, payment details and applicable VAT or currency treatment.

Those are Blue Canvas prices, not market averages. An audit price should not be mistaken for a complete build, software subscription or security certification. Check the live rate card and the written proposal before committing.

Ask every supplier to answer the same questions

  • What specific business task and users are included?
  • Which systems, data sources, integrations and environments are in scope?
  • Who will do the work, and who can answer technical questions?
  • Which output is a proposal, prototype, tested workflow or production release?
  • What normal and failure cases will be tested?
  • What remains under human approval?
  • Are licences, API usage, hosting and other recurring fees included?
  • What documentation, training, support and exit arrangements are included?
  • Which assumptions could change cost or delivery time?
  • How are changes requested, estimated and accepted?

Compare hourly, fixed-scope and retained work

Hourly work can suit investigation where the extent of the problem is not yet known. Agree a spending limit and a review point. Fixed-scope work can make the commitment clearer if deliverables, dependencies and acceptance criteria are specific. A retainer needs named responsibilities, support hours and exclusions.

Ask for a range of delivery assumptions where uncertainty is real. A fixed price does not remove the need to identify missing data, unavailable suppliers or internal decisions. Record what happens if one of those assumptions proves wrong.

Build your own cost-benefit worksheet

Start with the current task volume, time spent, error rate and cost of rework. Estimate how much review and exception handling will remain. Add build, licences, usage, support and staff training to the cost side. Keep cash savings separate from time that could be used for other work.

Fictional example: a team handles 40 requests a week. If a pilot saves 3 minutes of handling on each request but creates 30 minutes of extra review, the net time released is 90 minutes, not 120. This arithmetic illustrates a method; it is not a typical result or a client outcome.

Test whether the expected benefit survives lower volumes, more exceptions or higher usage costs. A useful proposal can acknowledge that a project may not be worth pursuing.

Evidence is more useful than a guaranteed return

Ask for relevant examples that the supplier has permission to show, and distinguish a demonstration from an independently operating system. For your own project, agree how the delivered version will be checked and who accepts it.

Treat unsupported guarantees, unexplained claims of universal savings and pressure to buy before the scope is clear as reasons to ask more questions. A supplier should be able to describe limitations and failure handling plainly.

Decide how much help you need

For a low-risk personal task, a short internal experiment may answer the question. Connected workflows, confidential data, customer actions or operational dependence usually need more structured design and review. Team size and turnover alone do not decide which route is appropriate.

Use our DIY-versus-consultancy decision guide to choose the next step, then compare any proposals on the same written basis.

For help scoping the work, see AI consultancy or book a free consultation. Agree the scope, responsibilities and written quote before work starts.

Read more

No items found.

Have a conversation with our specialists

Talk to Blue Canvas. Bring one task that takes too long.

In a free 15-minute call we will tell you whether AI can help, and what a sensible first step would cost.

Thanks. Your message has reached Blue Canvas.
Oops! Something went wrong while submitting the form.