AI HackathonsConsultingAbout
    Buyer guide

    How to choose an AI training provider in the UK

    Choose an AI training provider that gets your people practising on the AI tools you already pay for, using your own tasks, and that can show you who will deliver, how your data will be protected and how results will be measured. Brief two or three providers in the same way, ask each one for a named trainer, a sample agenda and a written list of what follow-up is included, then compare them on price per person and per day.

    Published by Day Seven, a Leeds-based AI training provider. It's a buyer's checklist, not an independent ranking: we compare types of provider, not named companies. Apply it to us as strictly as you would to anyone else.

    By Qasim Ahmed, founder of Day Seven · Updated

    Book a callJump to the checklistWhat AI training costs
    A facilitator leading a group learning session
    Illustrative image of group learning; not a documented Day Seven client session.

    In short

    • Judge providers on what your people will do in the session, not on the slide deck.
    • Train on the tool you already pay for: Microsoft Copilot, ChatGPT, Claude or Gemini.
    • Get the trainer's name, a sample agenda, the data rules and the follow-up in writing.
    • Agree how you'll measure results before you sign.
    • Compare quotes per person and per day, on an identical brief.

    Before you shop: write a one-page brief

    If every provider gets the same brief, you can compare like with like. Otherwise you end up comparing quotes for different pieces of work. Put these on one page:

    1. Who. Roles, headcount and how comfortable people are with AI today.
    2. Tools. Which AI tools you already license (Copilot, ChatGPT, Claude, Gemini or none yet), and what your policy allows.
    3. Work. Three to five repeated tasks that eat time, such as reports, emails, tickets, proposals and meeting notes.
    4. Constraints. Data rules, sites, shift patterns, accessibility needs and dates.
    5. Success. What you want to be different 30 and 90 days later, and how you'll know.

    Our AI training costs page has a buyer brief you can copy.

    The checklist: ten things to compare

    1. Practical vs theory

    People keep what they've done, not what they've watched. Ask what share of the session is participants working in the tool, and what each person will have produced by the end.

    • Good looks like: a run sheet with exercise blocks, and output people take back to their desks.
    • Ask: "How much of the day is people typing, and what will they have made?"

    2. Tool coverage

    Training should match the tool your organisation has bought. Copilot in Word, Excel, Outlook and Teams is different from ChatGPT or Claude in a browser, and Gemini lives inside Google Workspace. A vendor-neutral provider can help if you run more than one tool or haven't chosen yet. A vendor-specific one may go deeper on that single product.

    • Good looks like: exercises built for your licence level, so nobody learns features they can't use.
    • Ask: "Which tools and licence levels will you train on, and how do you check what we have?"

    3. Bespoke vs off-the-shelf

    Off-the-shelf content is quicker and cheaper to buy. Bespoke content uses your own documents and workflows, which is what changes how a team works. Ask how much is genuinely tailored, and when.

    • Good looks like: a scoping conversation before design, and draft exercises that name your workflows.
    • Ask: "Which parts are standard, and which will you build around our tasks?"

    4. Delivery format

    In person suits a team learning together. Live online suits teams spread over several sites. Self-paced suits the basics and new starters. Check group size, because it decides whether anyone gets individual help.

    • Good looks like: a stated cohort size, a plan for several cohorts and a straight answer on accessibility.
    • Ask: "How many people per session, how do you handle larger headcounts, and how are access needs met?"

    5. Who delivers

    Find out who will stand in front of your team, not who runs the company.

    • Good looks like: a named trainer whose delivery history is described, and a chance to speak to them before you book.
    • Ask: "Who's delivering, what have they delivered before, and will they be there on the day?"

    6. Follow-up support

    Habits form in the weeks after a session, not on the day itself. Ask what people keep and what support follows.

    • Good looks like: a prompt library or templates, an adoption plan, and a written split between what's included and what costs extra.
    • Ask: "What happens in the 30 days after the session, and what's included in the price?"

    7. Measuring ROI

    A provider with no view on measurement is telling you something. Agree on two or three real tasks, record a baseline before training, and measure again at around 30 days and at 60 to 90 days. Include the time spent checking AI output. Our guide to measuring AI training ROI sets out the formula and a template.

    • Good looks like: a measurement plan in the proposal, agreed before delivery.
    • Ask: "How will we know it worked, and what will you measure with us?"

    8. Data and security

    People need realistic material to practise on without putting confidential or personal data into the wrong tool. They also need to learn to check what the AI produces rather than trust it. UK GDPR still applies, and the ICO's guidance on AI and data protection is a good reference. If you don't yet have a policy, start with our employee AI policy guide.

    • Good looks like: a written data rule for the session, approved practice material, and a checking step inside every exercise.
    • Ask: "What data will people use in exercises, who approves it, and how is checking taught?"

    9. EU AI Act Article 4 (AI literacy)

    Article 4 of the EU AI Act requires providers and deployers of AI systems to take measures to support the AI literacy of their staff and of others using AI systems on their behalf. It has applied since 2 February 2025. The Digital Omnibus on AI (Regulation (EU) 2026/1744), in force from July 2026, amended it so that organisations don't have to guarantee any specific level of literacy for each individual. National authorities supervise it from 2 August 2026. It can apply to UK organisations that operate in the EU, or whose AI system output is used in the EU. This is not legal advice, so check your own position.

    • Good looks like: training matched to people's roles and the tools they use, with simple records of who was trained, on what and when.
    • Ask: "Can you give us attendance and content records we can keep as evidence of the measures we've taken?"

    10. Pricing models

    Prices aren't comparable until they're on the same basis. The common models are:

    • Per seat: open or public courses, and self-paced platforms (often per user per year).
    • Per day or per session: private workshops for one team, usually with a group-size cap.
    • Per programme: several sessions, coaching and follow-up over weeks or months.
    • Funded routes: apprenticeships and government-backed schemes, with eligibility rules and longer timescales.
    • Good looks like: a written price showing what's included (preparation, materials, travel, follow-up) and whether VAT is added.
    • Ask: "What's the cost per person, what's the group cap, and what isn't included?"

    Print-friendly checklist

    Print-friendly checklist for comparing AI training providers
    CriterionEvidence to requestProvider AProvider BProvider C
    Practical vs theoryRun sheet with exercise blocks
    Tool coverageTools and licence levels covered
    Bespoke vs off-the-shelfScoping step; draft exercises using your work
    Delivery formatCohort size; multi-site plan; accessibility
    Who deliversNamed trainer; call before booking
    Follow-up supportWhat's kept; included vs extra
    Measuring ROIBaseline and checkpoints in proposal
    Data and securityWritten data rule; checking step
    EU AI Act Article 4Training records you can keep
    Pricing modelPer person / per day; inclusions; VAT basis

    Questions to ask every provider

    About the training

    1. What will each person have produced by the end of the session?
    2. What share of the time is hands-on in the tool?
    3. Will the exercises use our tasks and our tool, or a standard example set?
    4. How do you handle mixed ability in the same room?

    About the people

    1. Who exactly will deliver it, and can we speak to them first?
    2. Can we speak to a client who has agreed to be contacted?

    About data and risk

    1. What data will people use in exercises, and who signs it off?
    2. How do you teach people to check AI output?
    3. What will you need from our IT team before the day?

    About results and support

    1. How will we measure whether it worked, and when?
    2. What support is there after the session, and what does it cost?
    3. What will we keep: prompts, templates, recordings, an adoption plan?

    About price and logistics

    1. What's the cost per person and per day, and what's the group cap?
    2. Are preparation, travel and materials included? Is VAT added?
    3. How soon can you deliver, and what's the cancellation policy?

    Red flags

    • No sample agenda, or an agenda made up of topics rather than exercises.
    • "We'll cover everything": every tool, every feature, in half a day.
    • Nobody can name the trainer, or the person on the call won't be the person delivering.
    • Real customer data in the exercises with no agreed rule on what's allowed.
    • Training on features you haven't licensed, such as paid Copilot features for staff who only have Copilot Chat.
    • No view on measurement, or success measured only by attendance and a feedback form.
    • Follow-up described vaguely ("ongoing support") without saying what's included.
    • Logos used as proof with no reference you can actually contact.
    • Badges with no explanation of who awarded them or what they assess.
    • Pressure to sign a long programme before a single pilot session.
    • Prices you can't compare: no per-person figure, no group cap, no VAT basis.

    Types of AI training provider compared

    These are general patterns, not a review of any company. Individual providers vary, so check them against the checklist above.

    Types of AI training provider compared
    Comparison pointLarge training companiesMicrosoft partners and IT resellersOnline course platformsFreelance trainersSpecialist AI training providers (like Day Seven)
    Typical strengthsBig catalogues, scheduled public dates, established admin and booking, often certificatesDeep Microsoft knowledge; can pair training with licensing and tenant configurationLow cost per person, available any time, scales to large numbersFlexible, often good value for small groups, direct contact with the trainerFocused on AI tools at work, usually tailored to your tasks, often cover several AI tools
    Watch-outsContent can be generic; trainer may vary by dateTraining may focus on Microsoft tools only; tool-led rather than task-ledCompletion and real use can be low without live support; little tailoringCapacity and cover if the trainer is ill; range of tools may be narrowSmaller teams mean less capacity and fewer public dates; check the trainer and references carefully
    Tool coverageBroad, by courseMostly Microsoft 365 Copilot and the Microsoft stackBroad but genericDepends on the personUsually several: Copilot, ChatGPT, Claude, Gemini
    TailoringLow to mediumMedium (for Microsoft environments)LowMedium to highHigh
    Usual pricing modelPer seat; private days on requestOften bundled with licensing or services projectsPer user licencePer day or per hourPer day, per seat or per programme
    Best forIndividuals needing a certificate; very large cataloguesOrganisations that also need Copilot licensing and tenant setupBroad awareness at scale; new startersSmall teams with a simple, defined needTeams from six people, ideally 10 or more, who want to change how specific work gets done

    Where Day Seven fits

    Day Seven is a Leeds-based AI training company for teams in UK organisations, from six people and ideally 10 or more. Here's how we answer the checklist, so you can hold us to it:

    • Practical: hands-on workshops where people work on their own tasks, around 16 people per workshop, with more sessions for larger groups.
    • Tools: vendor-neutral. We train on the tool you license: Microsoft Copilot, ChatGPT, Claude or Google Gemini.
    • Bespoke: a short scoping conversation comes first, then exercises built around each team's real use cases.
    • Delivery: at your offices anywhere in the UK, at a Leeds venue, or live online, usually two to three weeks from the first conversation.
    • Follow-up: teams leave with a shared prompt library and an adoption plan. Further coaching and check-ins can be scoped in, and the proposal says in writing what's included.
    • Measurement: we agree what to compare afterwards, such as time and output quality on a chosen task, plus confidence and adoption.
    • Copilot: we run readiness checks and training. Licensing and tenant/admin setup stay with your IT team or Microsoft partner.
    • Price: from £450 per seat, and from £6,950 per day for a private team workshop.
    • Beyond workshops: AI hackathons, including the seven-day Day Seven Sprint; training consulting on AI policy and adoption; and the self-paced Day Seven Academy.

    Day Seven has delivered AI enablement work for organisations including Velociti, Studiospace, Archer Hotel Capital and Yorkshire Cancer Research. This is company-level experience, not a claim that each organisation bought the service described on this page.

    Not the right fit? If you're learning on your own, see AI training for individuals. If you need an apprenticeship or a formal qualification, a funded provider or a large training company will suit you better, and we'll say so.

    Talk it through

    Book a call and we'll go through your brief together, even if you end up choosing someone else.

    Book a callEmail hello@dayseven.aiCall 07772 547459Or get a training proposal

    Not sure where your team stands? Try the free 3-minute AI readiness scorecard.

    Frequently asked questions

    Look for hands-on practice on your own tasks, training on the tools you already license, a named trainer you can speak to first, a clear rule on what data people may use, follow-up that is written into the proposal, and an agreed way of measuring results. A provider that can't show you a sample agenda with exercise blocks is probably selling a presentation, not training.

    No. Day Seven, an AI training provider, publishes this guide. We compare types of provider rather than naming or scoring other companies. Apply every criterion here to us as strictly as you apply it to anyone else.

    It depends on the pricing model. Open courses are usually priced per seat, private team workshops per day or per session, self-paced platforms per user licence, and longer programmes per engagement. Day Seven's training starts from £450 per seat, and private team workshops from £6,950 per day. To compare quotes fairly, work out the cost per person and check what follow-up is included.

    Not for training staff to use Copilot well. A Microsoft partner or your IT team is the right home for licensing, tenant configuration and permissions. A training provider should check that you're ready (who has which licence, what data Copilot can see, and what your policy allows) and then train people on real tasks. Some organisations use both.

    It can. Article 4 of the EU AI Act requires providers and deployers of AI systems to take measures to support the AI literacy of their staff. It has applied since 2 February 2025 and was amended in July 2026, so it no longer requires a specific level of literacy for each person. It can reach UK organisations that operate in the EU or whose AI output is used there. Take legal advice on your own position. Documented, role-relevant training is a sensible way to show that you have taken measures.

    Off-the-shelf courses suit individuals and broad awareness, and they are quick to buy. Bespoke training suits teams that need to change how specific work gets done, because the exercises use their own documents, tools and rules. Many organisations combine the two: self-paced courses for the basics, plus a tailored live session for each team.

    Accreditation can help, but check who awards it and what it covers. Some badges assess a course's paperwork, not the trainer in the room. A named trainer, a sample agenda, a reference you're allowed to contact and an agreed measurement plan usually tell you more.

    Before the training, choose two or three real tasks and record how long each takes, how often it happens and how often it needs rework. Measure again at about 30 days and at 60 to 90 days, including the time people spend checking AI output. Track adoption and confidence alongside time saved. Attendance and satisfaction scores alone don't tell you much.

    Sources