Author: AP

  • Write Down How Your Business Works in a Week, Using AI

    In most owner-managed businesses, the way things are done lives in people’s heads, mostly the owner’s. That works until someone is off sick, someone leaves, or you try to take a week off. Then the business wobbles.

    Writing it all down sounds like a six-month project nobody has time for. With AI doing the typing, it doesn’t have to be. Here’s how to get your key processes written down in a week.

    Day 1: Pick the ten jobs that matter most

    Don’t try to document everything. List the ten jobs that would cause the most trouble if they went wrong, or if the person who does them was away. For most businesses that includes taking an order, raising an invoice, chasing payment, ordering stock, setting up a new customer and the month-end routine.

    Days 2 to 4: Talk, don’t type

    For each job, ask the person who does it to talk it through for ten minutes, recorded on a phone or on a video call with transcription switched on. They should explain it as if they were training someone new: what starts the job, each step, the systems they use, and what usually goes wrong.

    Then give the transcript to your AI assistant.

    “Here is a transcript of someone explaining how they process a new customer order. Turn it into a clear step-by-step process with numbered steps, the system used at each step, a short checklist at the end, and a list of questions where the explanation was unclear.”

    The list of unclear questions is the most valuable part. It shows you exactly where the business depends on someone’s judgement or memory.

    Day 5: Check, fix and keep it in one place

    Give each written process back to the person who does the job and ask them to correct it. Then store everything in one shared place your team already uses, such as a shared drive or SharePoint, with a simple index page.

    Keeping it up to date

    A process document is only useful if it’s right. Agree one owner for each process and a simple rule: if you change how you do it, you update the page. Once a quarter, ask each owner to check theirs. AI helps here too: paste in the old version and a description of what has changed, and it will rewrite it for you.

    What you have at the end of the week

    Ten written processes, a list of the places where the business relies on one person’s memory, and a much easier week off. It’s also the groundwork for automation, because you can’t automate a job nobody has written down.

    A typical example

    A business with 20 staff spends about an hour per job recording and checking ten processes, plus a few hours of the owner’s time pulling it together. That’s roughly two days of effort in total, spread across a week. Swap in your own numbers.

    Writing the processes down is the easy part. Getting the team to follow them, and fixing the ones that don’t work, is where most businesses stall. That’s the part we help with. We’re not IT people. We’ve sat in the MD’s chair and built the systems behind it, and we’ll talk to you in plain business terms.

    Growing pains like this are normal in a business that has grown quickly, and they’re fixable. Take the free Growing Pains Check to see where your business stands in three minutes, or book a free call about a Growing Pains Diagnostic: half a day with you and your business, then a written 90-day plan with the three changes that matter most.

    See how the workshops work →

    Or book a free 30-minute call with us →

  • 7 Admin Jobs You Can Hand to AI This Month (No Technical Skills Needed)

    In most growing businesses the admin grew faster than the team, and too much of it still lands on the owner’s desk. These are the jobs AI can take off the pile first, without anyone needing to be technical.

    Most small business owners don’t need an “AI strategy”. They need their evenings back.

    According to the Office for National Statistics, around a third of UK businesses with 10 or more staff now say they use AI. For the rest, the biggest things holding them back are not having the know-how and not knowing where to start.

    So let’s start. Here are seven everyday admin jobs you can hand to Claude this month, each with a prompt you can copy straight in. A quick ground rule first: keep sensitive personal data out, and always read what Claude gives you before you use it.

    We use Claude in the examples below because it’s the AI assistant we use most day to day. We’ll set you up with whichever AI tool suits your business.

    1. Getting through your inbox faster

    Paste in that sprawling email thread and ask:

    “Summarise this thread in three bullets, tell me what I’m being asked to do, and draft a short reply that says yes to X and asks for Y.”

    If you connect Claude to your Outlook or Gmail, you can skip the copying and pasting and just ask it to find the thread.

    2. Turning meeting notes into actual actions

    Paste in your scribbled notes and ask:

    “Turn these notes into a list of actions, each with an owner and a deadline, then write a short follow-up email to everyone who attended.”

    3. Writing quotes and proposals

    Upload an old proposal you were happy with, plus a few notes on the new job, and ask:

    “Using the style and structure of the attached proposal, draft a new one for [client] based on these notes. Keep it under two pages and save it as a Word document.”

    4. Keeping tabs on the competition

    Checking competitors’ websites and prices is exactly the kind of repetitive job that never gets done. Once a month, ask Claude:

    “Search the web for anything [competitor names] have announced or changed in the last month, summarise it, and suggest one thing we should respond to.”

    5. Job adverts and first-draft policies

    “Draft a job advert for a part-time bookkeeper, 20 hours a week, based in [town]. Friendly tone, clear list of duties, no jargon.”

    Anything with legal weight, like contracts or disciplinary policies, still needs checking by someone qualified.

    6. Getting five pieces of content out of one

    “Turn this blog post into three LinkedIn posts, one customer email and a short paragraph for our newsletter.”

    7. Writing down how things are done

    Type or dictate rough notes on how you do a task, however messy, then ask:

    “Turn this into a step-by-step procedure a new starter could follow, with a checklist at the end.”

    It’s the easiest way we know to get processes out of someone’s head and onto paper.

    Next step: make it a habit, not a one-off

    Once a prompt works, save it. Create a Project in Claude for each regular job, put your instructions in once, and your whole team can use it the same way every time.

    You can also connect Claude to your email, calendar and files, so it works with your real information instead of whatever you paste in. And on Claude’s paid plans, Claude Cowork can run regular jobs on a schedule, like a weekly summary of your inbox waiting for you on a Monday morning.

    What it adds up to

    Here’s a typical example. If each of these seven jobs saves you 30 minutes a week, that’s 3.5 hours a week. Over a year it’s around 160 hours, or four working weeks.

    Curious how a managing director actually uses this day to day? Have a read of What I Actually Use Claude For.

    Your one-week challenge

    1. Pick two jobs from the list.
    2. Try each prompt three times this week.
    3. Jot down roughly how much time it saved.
    4. Save the prompts that worked in a Claude Project for the team.

    Want the whole team doing this?

    Our one-day workshops take your team through exactly this, using your own real tasks. Take the 3-minute AI Readiness Check to see where you stand.

    Source: Office for National Statistics, “Artificial intelligence in UK businesses: 2023 to 2026”, Business Insights and Conditions Survey, June 2026.

    We’re not IT people. We’re business leaders who’ve put AI to work across sales, finance, operations and marketing. We can set this up for you, train your team and hand it over, or keep running it for you each month.

    Growing pains like this are normal in a business that has grown quickly, and they’re fixable. Take the free Growing Pains Check to see where your business stands in three minutes, or book a free call about a Growing Pains Diagnostic: half a day with you and your business, then a written 90-day plan with the three changes that matter most.

    See how the workshops work →

    Or book a free 30-minute call with us →

  • Which Spreadsheet Is Right? Getting to One Set of Numbers You Trust

    Ask three people in a growing business for last month’s sales and you’ll often get three different answers. One came from the accounts, one from the sales spreadsheet and one from someone’s memory. Each is right in its own way, and none of them helps you make a decision.

    One version of the numbers is one of the first things we fix with owner-managed businesses, because almost everything else depends on it. Here’s how to get there, with AI doing most of the legwork.

    Why the numbers drift apart

    It’s rarely anyone’s fault. Spreadsheets get copied and adapted, each department builds its own, and definitions slip. Is “sales” before or after credit notes? Does “this month” mean the invoice date or the order date? Over five or ten years you end up with several versions of the truth.

    Step 1: List the numbers that actually matter

    You don’t need fifty measures. Most owner-MDs run the business on six to ten: sales, gross margin, overheads, cash in the bank, money owed to you, money you owe, and a few that are specific to your trade. Write your list down.

    Step 2: Agree one definition and one source for each

    For each number, agree exactly what it means and where it comes from. Usually that’s your accounts software for anything financial and your sales or stock system for the rest. Put it in one table: the number, the definition, the source, and who is responsible for it being right.

    AI is useful for the first draft. Give it your list and ask it to suggest definitions and spot where people might be counting things differently.

    “Here are the ten numbers I use to run my business. For each one, write a clear one-line definition, suggest the most reliable source, and list the common ways different people end up calculating it differently.”

    Step 3: Retire the copies

    This is the hard bit, and it’s a people job more than a software one. Once the table is agreed, old spreadsheets get archived, not updated. If someone needs a number, it comes from the agreed source.

    Step 4: Let AI build the weekly summary

    Now the useful part. Each week, export the agreed figures and let your AI assistant turn them into the same one-page summary: the numbers, how they compare with last week and last year, and anything unusual worth a question. Claude, ChatGPT and Microsoft Copilot can all read a spreadsheet export and do this, as long as you give them the same instructions each time.

    “Here is this week’s export. Using the definitions below, produce a one-page summary with this week, last week and the same week last year, a short chart, and three things I should ask about.”

    Always check the summary against the source before it goes to anyone. AI is very good at this job, but a person should check the work and make the decisions. Use a business account for your AI assistant and check its data settings before sharing financial details.

    What changes when the numbers agree

    Meetings get shorter, because nobody argues about which figure is right. Decisions get quicker, because the information is already there. And you can finally see whether turnover growth is turning into profit.

    If you just want the weekly report built for you, our Sales Report Setup does that for a fixed price. Getting the whole business onto one version of the numbers is part of what we do in MD Advisory. That’s the part we help with. We’re not IT people. We’ve sat in the MD’s chair and built the systems behind it, and we’ll talk to you in plain business terms.

    Growing pains like this are normal in a business that has grown quickly, and they’re fixable. Take the free Growing Pains Check to see where your business stands in three minutes, or book a free call about a Growing Pains Diagnostic: half a day with you and your business, then a written 90-day plan with the three changes that matter most.

    See how the workshops work →

    Or book a free 30-minute call with us →

  • Build a CRM With AI: What I Learned Doing It for a Sales Team

    You can build a CRM with AI using the spreadsheets and diaries your sales team already keeps, and I did exactly that for a field sales team this week.

    We built this in Claude, the AI assistant we use most day to day. We’ll set you up with whichever AI tool suits your business.

    How to Build a CRM With AI From What You Already Have

    Most small sales teams don’t need a new system. They need the three they’ve already got to talk to each other.

    Ours had a pipeline tracker in Excel, a weekly activity sheet per rep and four Outlook diaries. Every meeting was recorded somewhere. Nobody could see them in one place, and follow-ups lived in people’s heads.

    So I asked Claude to read all three and build one page from them. The first load pulled in 41 customer meetings. For each one it records:

    • The business, the person and their job title
    • Address, phone and email
    • Sector, and whether they’re a new or existing customer
    • Meeting length, the objective and the outcome
    • The pipeline value and stage, where there is one

    It also shows the meetings on a map. You pick a centre, either head office or a rep’s patch, and drag a radius from 10 to 250 miles. Seeing who’s travelling where tells you more than any activity report.

    If your sales activity is spread across Excel files and Outlook like this, we can map out what a first version would look like on a 30-minute call. It’s usually quicker than people expect.

    Book a 30-Min Discussion

    Build the Follow-Up In, Not On Top

    A meeting without a follow-up is a nice coffee. I wanted the CRM to nag, politely.

    Every meeting gets two dates. The follow-up email is due by close of business on the Friday of the meeting week. The chaser is due the Friday after. Claude sends each rep a calendar invite for both, so the reminder sits in the diary they already live in.

    Each meeting also carries a short list of prompts the rep ticks off:

    • Add or update the pipeline tracker
    • Connect with the contact on LinkedIn
    • Ask them to follow our company page
    • Ask for a review (existing customers only)

    That last rule matters. Asking a new prospect to rate you after one meeting is a quick way to stay a prospect.

    Keep the AI CRM One-Way

    This was the first real decision, and I’d make it the same way again.

    The obvious question was whether a rep could change a pipeline value in the CRM and have it flow back into Excel. It could be built. I said no.

    Excel and Outlook stay the master copy. Claude refreshes the CRM every weekday morning. If a number’s wrong, it gets fixed in the tracker, and the next morning’s refresh picks it up. One source of truth beats two that argue.

    The refresh also showed me problems I hadn’t spotted. Deals had weighted values against a £0 total. One rep’s activity tab was weeks out of date. Another’s sheet had the wrong month in the heading. None of it was hidden, it just wasn’t anyone’s job to look.

    What Tripped Us Up

    Most of the time didn’t go on building. It went on access.

    Claude could read two of the four diaries straight away. The other two needed sharing at the “Can view all details” calendar permission in Outlook. Seeing a colleague’s diary yourself isn’t the same as an AI tool being allowed to read it.

    Claude also can’t write into someone else’s diary, even with editor access. The workaround is invites from mine, which land in theirs automatically.

    When I shared the page, the reps couldn’t save anything at first. The usual causes are a public link being switched on, or someone opening it while signed in with the wrong account. Check both before you blame the tool.

    Customer contact details need care too. If reps will email people from it, the ICO’s guide to business-to-business marketing rules is short and worth reading.

    Where It Goes Next

    Activity is half the picture. The other half is sales, the same numbers behind a weekly sales report.

    The next step is a daily export of invoiced sales and margin by customer from our reporting tool into SharePoint. Power Automate’s Power BI connector can run that query on a schedule, and the CRM reads the file each morning.

    That gives two views. A management dashboard shows each rep’s sales, margin and meetings against their monthly target. Each rep gets a private page with only their own customers. League tables are for managers, not the shared screen.

    Andrew’s Final Thoughts

    An AI CRM won’t replace Salesforce for a 50-person sales floor. For a team of four with good spreadsheets and no time, it’s the fastest way I’ve found to see who met whom, what was promised and what’s overdue. Start with one-way data and the follow-up dates. Not sure your data is ready? The free AI Readiness Check takes a few minutes. If you’d like it built for you, have a look at our fixed-price setups.

    Build a CRM With AI FAQs

    Can you build a CRM with AI without a developer?

    Yes. Claude can read your spreadsheets, diaries and email through connectors and build a working CRM page. You need to know what fields you want and where the data lives. A developer helps for heavy customisation, but a small team can get started without one.

    Is a homemade AI CRM a replacement for Salesforce or HubSpot?

    Not for large teams. It suits smaller teams who already keep tidy sheets and want one view of meetings, follow-ups and pipeline. If you need marketing automation, complex permissions or hundreds of users, a dedicated CRM is still the better choice.

    Should an AI-built CRM sync both ways with Excel?

    I’d start one-way. Keep the spreadsheet as the master and update the CRM from it each day. Two-way sync means two places a number can change, and you’ll spend time working out which one is right. Add write-back later, if the team actually needs it.

    What data does a small sales team CRM need?

    The business, contact name and job title, address, phone, email, sector, customer type, meeting objective, outcome and pipeline value. Add follow-up dates and a few prompts, such as updating the pipeline or requesting a review, and you’ve covered most of what matters.

    How do I get sales reps to actually use the CRM?

    Make it come to them. Follow-up reminders go into their own diary, the CRM fills itself from sheets they already complete, and they only tick boxes or add a missed meeting. For more ideas on getting a team using AI, see our workshops.

    If you would rather have a simple client tracker built for you, see our fixed-price Client Tracker Setup.

    Growing pains like this are normal in a business that has grown quickly, and they’re fixable. Take the free Growing Pains Check to see where your business stands in three minutes, or book a free call about a Growing Pains Diagnostic: half a day with you and your business, then a written 90-day plan with the three changes that matter most.

  • Every Decision Still Comes Back to You: How to Stop Being the Bottleneck

    If you run a growing business, you probably know the feeling. You walk back in after a morning out and there’s a queue: three people at your door, a dozen emails marked urgent, and a quote that has been waiting since yesterday for your say-so.

    It feels like proof you’re needed. In reality it’s often the biggest thing holding the business back, because the business can only move as fast as you can answer.

    Why it happens

    In the early years, the owner making every call is a strength. You know every customer, every price and every supplier. The trouble is that the habit outlives the business it suited. Somewhere around 10 to 20 staff, the number of decisions grows faster than one person can handle, but nobody ever sits down and moves them.

    Step 1: Write down what comes to you for a week

    For one week, keep a simple list of every question or decision that lands with you. Don’t sort it, just capture it. Voice notes are fine.

    At the end of the week, paste the list into your AI assistant and ask it to sort them.

    “Here is every decision that came to me this week. Group them into types, count how many of each, and for each type say whether it needs me, could be decided by someone else with clear rules, or could be avoided altogether.”

    Most owners are surprised by the answer. A large share of what reaches them is the same handful of questions, asked again and again.

    Step 2: Turn the repeat questions into rules

    For each repeat decision, write down the rule you already use in your head. When can a discount be given, and how much? Who can approve overtime? When does a complaint need you, and when can it be settled with a credit?

    This is where AI saves real time. Talk through how you decide, and let it write the rule up as a short, clear guide.

    “I’m going to explain how I decide whether to give a customer a discount. Turn what I say into a simple one-page rule a sales administrator could follow, with examples, and list any situations where they should still check with me.”

    Step 3: Give each type of decision an owner

    A rule without an owner just moves the queue somewhere else. Name one person for each type of decision, tell the team, and agree how they’ll let you know what they decided. A short weekly summary is usually enough.

    Step 4: Make the information come to you, not the questions

    Once others are deciding, you need to see what’s happening without being asked. This is where a simple weekly report earns its keep: sales, margin, cash, complaints and anything unusual, on one page, built the same way every week. AI can pull that together from your exports so nobody spends Monday morning building it.

    What to keep for yourself

    Not everything should be handed over. Keep the decisions that shape the business: pricing strategy, key hires, your biggest customers and where the money goes. The aim isn’t to step away. It’s to spend your time on the decisions only you can make.

    A typical example

    Take an owner who spends an hour a day answering repeat questions from the team. That’s five hours a week. If clear rules and named owners take half of those questions off their desk, that’s two and a half hours a week back, or around 120 hours a year. Swap in your own numbers.

    Moving decisions is simple to describe and hard to stick to, especially when it’s your own business. That’s the part we help with. We’re not IT people. We’ve sat in the MD’s chair and built the systems behind it, and we’ll talk to you in plain business terms.

    Growing pains like this are normal in a business that has grown quickly, and they’re fixable. Take the free Growing Pains Check to see where your business stands in three minutes, or book a free call about a Growing Pains Diagnostic: half a day with you and your business, then a written 90-day plan with the three changes that matter most.

    See how the workshops work →

    Or book a free 30-minute call with us →

  • Five Signs Your Business Has Outgrown Its Teenage Years (and What AI Can Fix This Month)

    Most owner-managed businesses hit the same wall somewhere between five and fifteen years in. Sales and headcount have grown, but the business still runs on the owner’s memory, goodwill and long hours. We call it the teenage years: plenty of energy, not much structure, and growing pains everywhere.

    None of that is failure. It’s a sign the business has grown faster than the way it’s run. Here are the five signs we see most, and for each one, what AI can realistically fix this month and what needs more than software.

    1. Every decision still comes back to you

    Pricing questions, customer complaints, which supplier to use, whether someone can have Friday off. If your phone rings for all of it, the business can’t grow past the hours in your day.

    What AI can fix this month: the groundwork. Most decisions that reach you need the same information pulled together first. An AI assistant can turn a messy email chain into a short summary with the options laid out, so you decide in two minutes instead of twenty.

    “Here’s an email thread about a customer complaint. Summarise what happened, what the customer wants, and give me three options with the pros and cons of each.”

    What needs more: deciding which decisions should never reach you in the first place, and who owns them instead.

    2. Nobody’s sure which spreadsheet is right

    The sales figure in the board pack doesn’t match the one in the accounts, and the stock sheet has three versions with “FINAL” in the name. Meetings turn into arguments about the numbers instead of what to do about them.

    What AI can fix this month: the weekly pull-together. ChatGPT, Microsoft Copilot and Claude can all read a sales export and turn it into a summary, a chart and a list of anything unusual, the same way every week.

    What needs more: agreeing one source for each number and who keeps it right.

    3. Turnover’s growing, but profit isn’t keeping up

    More customers, more staff, more hours, and somehow the bottom line hasn’t moved. Usually the margin is leaking somewhere nobody is looking: discounts, a handful of unprofitable customers, or work being done twice.

    What AI can fix this month: finding where to look. Give an AI assistant your sales by customer and product, with costs where you have them, and ask it to rank them by margin and flag the outliers. It won’t be perfect, but it will tell you which questions to ask your accountant.

    What needs more: acting on it. Repricing a customer or dropping a product line is an MD decision, not a software one.

    4. Good people leave because nobody told them what good looks like

    In a small business, people learn the job by watching the owner. Once you have 15 or 20 staff, that stops working. Good people get frustrated, and the ones who stay keep coming back to you.

    What AI can fix this month: the paperwork nobody has time to write. An AI assistant can turn a ten-minute voice note about a role into a clear role description, the five things that matter most in it, and a simple monthly one-to-one template.

    What needs more: actually having the one-to-ones, every month, with everyone.

    5. A week off feels impossible

    This is the one owners mention last but feel most. If the business wobbles when you’re away for a week, it still depends on you for something that should be written down or handed over.

    What AI can fix this month: the handover pack. Talk an AI assistant through what happens in a normal week and what only you know, and ask it to turn that into a checklist someone else could follow.

    What needs more: trusting someone else to use it, and resisting the urge to check your email.

    Where to start

    If two or more of these sound familiar, start with the one that costs you the most time each week. Pick one AI fix from the list above and use it on real work this week. Use a business account for your AI assistant and check its data settings before sharing customer or financial details.

    Then look at the bigger question: what needs to change so the business runs properly without you in every room. That’s the part we help with. We’re not IT people. We’ve sat in the MD’s chair and built the systems behind it, and we’ll talk to you in plain business terms.

    Growing pains like this are normal in a business that has grown quickly, and they’re fixable. Take the free Growing Pains Check to see where your business stands in three minutes, or book a free call about a Growing Pains Diagnostic: half a day with you and your business, then a written 90-day plan with the three changes that matter most.

    See how the workshops work →

    Or book a free 30-minute call with us →

  • Spot the Customers Going Quiet (and Make Sense of Your Reviews) With AI

    When a business grows, the owner stops hearing every customer conversation. That’s when quiet customers slip away unnoticed. AI can now spot the warning signs for you, in the data you already have.

    Most small businesses don’t lose customers in a blazing row. They lose them quietly.

    The regular who ordered every month hasn’t ordered since June. The client who used to reply within the hour now takes a week. Nobody notices until the renewal doesn’t arrive, and by then they’ve found someone else.

    The warning signs are nearly always sitting in your systems already. The trouble is that nobody has time to go and look. Claude can do the looking for you.

    We use Claude in the examples below because it’s the AI assistant we use most day to day. We’ll set you up with whichever AI tool suits your business.

    Part 1: Spotting customers who are going quiet

    Decide what “going quiet” means for you. Keep it simple. For example, no order in 60 days from someone who normally orders monthly, spend down more than 30% on the same quarter last year, or no reply to your last two emails.

    Give Claude the numbers. Export a list of customers from your accounts or order system with their last order date and spend by quarter. Nearly every system can do this as a spreadsheet. Create a Project in Claude called “Customer check” and put your “going quiet” rules in the instructions.

    Let Claude do the scanning. Drop the export into a chat and ask:

    “Using our rules, list every customer who looks like they’re going quiet, starting with the most valuable. For each one, suggest a friendly reason to get in touch and draft a short, personal check-in email.”

    If you connect Claude to your Outlook or Gmail, it can also search your recent emails with a customer, so you know where things stand before you pick up the phone.

    Turn it into a habit. Fifteen minutes every Monday. Run the list, pick five customers, make five calls. That’s honestly it.

    Part 2: Making sense of your reviews

    Google reviews, survey replies, the odd “just wanted to say” email. Most businesses collect far more feedback than they ever read properly.

    Copy a few months of it into Claude (a straight copy and paste is fine) and ask:

    “Group this customer feedback into themes. For each one, tell me how often it comes up, give me one example quote, and tell me whether it’s getting better or worse.”

    In a few minutes you’ll see patterns that would take an afternoon to find by hand, like delivery times coming up in one review in five, or the same member of staff being praised by name again and again.

    Claude can also draft replies to reviews: a proper thank you for the good ones, and a calm, helpful first draft for the ones that sting. Just make sure a real person reads and tweaks every reply before it goes live.

    What it’s worth

    Here’s a typical example. Say you have 200 active customers, each worth £2,000 a year. If a Monday check helps you keep just three who would otherwise have drifted off, that’s £6,000 a year protected, for 15 minutes a week.

    Keeping it safe

    Only share the customer data Claude actually needs. Names, order dates and spend usually do the job, and you rarely need addresses or bank details. Use a Claude business plan, or check your privacy settings, and make sure your own privacy notice covers how you use customer data.

    Starter checklist

    1. Write down your own “going quiet” rules.
    2. Export last order date and spend for every customer.
    3. Set up a “Customer check” Project in Claude with your rules.
    4. Run the prompt above and sanity check the list.
    5. Book a 15-minute slot every Monday.

    Want help making it a routine?

    We cover customer tracking in our Sales deep dive workshop and can set the whole thing up in Claude for you through AI Implementation. Take the 3-minute AI Readiness Check to see where you stand.

    We’re not IT people. We’re business leaders who’ve put AI to work across sales, finance, operations and marketing. We can set this up for you, train your team and hand it over, or keep running it for you each month.

    Growing pains like this are normal in a business that has grown quickly, and they’re fixable. Take the free Growing Pains Check to see where your business stands in three minutes, or book a free call about a Growing Pains Diagnostic: half a day with you and your business, then a written 90-day plan with the three changes that matter most.

    See how the workshops work →

    Or book a free 30-minute call with us →

  • Claude vs ChatGPT vs Copilot: Which AI Should an MD Choose? (2026)

    Every MD we speak to asks the same thing: which AI should we actually use? Claude, ChatGPT or Microsoft Copilot?

    The honest answer is that all three are good, and the best choice depends far more on how your business works than on which is “smartest” this month. Here’s a plain-English comparison of what each costs, how they treat your data, and where each one fits, using the providers’ own pages (checked 7 October 2026).

    The short answer

    If your business lives in Microsoft 365 (Outlook, Word, Excel, Teams), look hard at Copilot, because it works inside those apps.
    If most of your AI use will be writing, thinking things through, and working with long documents and spreadsheets, Claude is the one we use most ourselves.
    If you want a broad all-rounder with a large range of add-ons, ChatGPT is a strong choice.

    That’s our view from using them in real businesses, not a lab test. The right answer for you comes from trying them on your own work.

    What they cost for a business

    Claude Team: $20 per person a month billed annually, or $25 monthly, for 2 to 150 people (Claude pricing).
    ChatGPT Business: $20 per person a month billed annually, or $25 monthly, minimum two people (ChatGPT Business).
    Microsoft 365 Copilot Business: an add-on for existing Microsoft 365 business customers at £19.32 per person a month on a monthly plan, or £13.80 on an annual plan under a promotion that runs to 31 December 2026 for the first year (Microsoft 365 Copilot).

    Claude and ChatGPT list prices in US dollars, so the amount you pay in pounds may differ. All prices exclude VAT. Each provider also has cheaper personal plans, but for a team we’d always start on a business plan.

    What happens to your business data

    This matters more than any feature. On their business plans, both Claude and ChatGPT say they don’t train their models on your content by default (Claude pricing; ChatGPT Business). On the free and personal plans, both list model training as something you can opt out of (ChatGPT pricing), so check the settings before anyone uses them for work.

    For Copilot, check Microsoft’s terms for your specific plan with whoever looks after your Microsoft 365 account.

    Where each one fits into your day

    Copilot works inside Word, Excel, PowerPoint, Outlook and Teams. Microsoft also says Copilot Chat is available at no extra cost to users with an eligible Microsoft 365 subscription, so it’s worth checking what you already have (Microsoft 365 Copilot).
    Claude includes Claude for Microsoft 365, covering Outlook, on its paid plans, alongside Projects for keeping work on one job together (Claude pricing).
    ChatGPT lists Projects, scheduled tasks and custom GPTs on its Plus plan and above (ChatGPT pricing).

    How to choose, in a month

    1. Pick two jobs your team does every week, such as a report and a customer email
    2. Try two of the tools on exactly those jobs for two weeks each
    3. Use business plans, and agree what data can and can’t go in before you start
    4. Ask the people doing the work which one saved them more time
    5. Choose one, set it up properly and train the team on it

    The worst outcome is everyone using a different free tool with nobody knowing what’s been shared. Picking one and setting rules matters more than picking the “best” one.

    Want help choosing?

    We’re not IT people. We’re business leaders who’ve put AI to work across sales, finance, operations and marketing. We use Claude most day to day, but we’ll set you up with whichever AI tool suits your business and train your team on it. See what it costs, or take the 3-minute AI Readiness Check to see where AI could save you the most time.

    Growing pains like this are normal in a business that has grown quickly, and they’re fixable. Take the free Growing Pains Check to see where your business stands in three minutes, or book a free call about a Growing Pains Diagnostic: half a day with you and your business, then a written 90-day plan with the three changes that matter most.

    See how the workshops work →

    Or book a free 30-minute call with us →

  • What Does AI Training Cost a UK Small Business? (2026 Prices)

    If you run a small business, you’ve probably been told your team needs “AI training”. The next question is always the same: what should it cost?

    The short answer: anywhere from free to around £3,500 a day, depending on whether you want general awareness or training built around your own business. Here’s what the market charges in 2026, what you’re actually paying for, and how to tell if it’s worth it.

    What AI training costs in the UK in 2026

    Free: the government’s AI Skills Boost programme offers free short courses for every UK adult, from about 20 minutes to a day, focused on everyday work tasks (Enterprise Nation, January 2026).
    £10 to £50 per person a month: online courses on the big learning platforms (Elansio).
    £400 to £900: a half-day workshop from a freelance trainer (LogicRoad).
    £980 to £1,950: a half-day workshop from a specialist AI trainer (LogicRoad).
    £1,800 to £3,500: a full-day workshop from a specialist (LogicRoad).
    £750 to £1,500 a day: specialist AI consultancies, against £2,000 to £5,000 a day for the big consultancies (Elansio).

    These are published ranges from providers’ own pricing guides, so treat them as a guide and always ask for a written quote.

    What you’re actually paying for

    Free and online courses are a sensible place to start. They teach what AI is and how to write a decent request. What they can’t do is look at your business.

    The difference with paid, in-person training is that it should be built around your own work: your invoices, your reports, your customer emails. People leave having done a real job with AI, not just watched a demo. Good training also sets clear rules on what data goes in and how answers get checked, and someone checks in afterwards to make sure it’s sticking.

    Per person or per session?

    Watch how the price is quoted. Some providers charge per person, which gets expensive fast as your team grows. Others charge per session, so the price is the same for 8 people or 20. For a small business, a per-session price is usually better value.

    How to tell if it’s worth it: a typical example

    Say 10 people each save just 1 hour a week after a workshop. That’s 10 hours a week, worth around £200 a week at £20 an hour, or about £10,400 a year. A £2,000 workshop would pay for itself in about 10 weeks. This is a typical example, so swap in your own numbers.

    Questions to ask any AI training provider

    1. Will you use our own tasks and documents, or generic examples?
    2. Is the price per person or per session, and is VAT included?
    3. Who actually delivers it, and have they run a business?
    4. Do you cover safe use: what data goes in, and how outputs are checked?
    5. What happens afterwards? Is there a follow-up?
    6. Which AI tools do you cover, and will you help us choose?

    What our workshops cost

    Our prices are on our pricing page, and every workshop is priced per session, not per person: a 1:1 with the MD from £1,000 + VAT, a Senior Leadership Team session from £2,000 + VAT, and a Department Deep Dive from £2,500 + VAT. Each one uses your own work and includes a 30-day check-in. If you’d rather we simply set AI up for one job, our fixed-price setups start at £595 + VAT.

    Start free

    Not sure you need paid training yet? Try a free government course, then take our 3-minute AI Readiness Check to see where AI could save your business the most time.

    We’re not IT people. We’re business leaders who’ve put AI to work across sales, finance, operations and marketing, and we’ll talk to you in plain business terms. See how our workshops work.

    Growing pains like this are normal in a business that has grown quickly, and they’re fixable. Take the free Growing Pains Check to see where your business stands in three minutes, or book a free call about a Growing Pains Diagnostic: half a day with you and your business, then a written 90-day plan with the three changes that matter most.

    See how the workshops work →

    Or book a free 30-minute call with us →

  • Before You Pay for a New Website: What AI Can Fix First

    In many growing businesses the marketing still depends on the owner, and the website shows it. Before you spend on a new one, be clear what it needs to do for the business.

    Most small business websites were built once and then left. The words are out of date, the services have changed, and updating anything means emailing someone and waiting. Sooner or later the answer seems to be a brand new website.

    That can be the right call. A UK guide verified in August 2026 puts a five to ten page small business site at £1,500 to £3,000 from a freelancer and £2,000 to £5,000 from a small agency (ExpertSure). Before you spend that, it’s worth knowing that AI can fix a lot of what’s wrong with most websites in an afternoon.

    What a small business website actually needs to do

    1. Say clearly what you do, and who for, in the first few seconds
    2. Show proof: examples, reviews, the kind of customers you work with
    3. Explain each service in plain English
    4. Give a clear next step on every page
    5. Be found on Google for what you do and where you work
    6. Stay up to date

    Most tired websites fail on the words and the upkeep, not the design. That’s exactly where AI helps.

    1. Rewrite your pages in plain English

    Copy the text from your home page into your AI assistant:

    “This is my website’s home page. Rewrite it so a busy small business owner understands in ten seconds what we do, who it’s for and what to do next. Keep it plain British English, no jargon, and suggest a clear headline and one call to action.”

    You still decide what’s true and what stays. But a blank page or a muddled one stops being the problem.

    2. Help Google understand what you do

    Every page has a title and a short description that Google shows in its results. Ask for both:

    “For each of these pages, write a page title under 60 characters and a description under 155 characters that says what we do and where we work. Use the words a customer would search for.”

    3. Keep it fresh without the hassle

    Give your AI assistant notes from a recent job, a customer question you get every week, or a new service, and ask for a short news post, a case study (with the customer’s permission) or a new FAQ. Ten minutes of your time becomes a page that keeps the site alive.

    4. See your site through a customer’s eyes

    Paste in the text of a page and ask:

    “Read this as a potential customer who has never heard of us. What’s unclear, what’s missing, and what would stop you getting in touch?”

    It’s a quick, honest second opinion, and it often finds the gaps people inside the business can’t see.

    When you really do need a new website

    AI won’t fix everything. It’s time for a new site if you can’t update it yourself at all, it looks broken on a phone, it’s painfully slow, or it’s built on something nobody supports any more. The good news is that AI makes a new site quicker and cheaper to build too.

    What it saves: a typical example

    The same guide puts ad hoc content updates at £50 to £100 an hour from an agency. If you need two hours of changes a month, that’s £100 to £200 a month. Doing the same updates yourself with AI might take an hour, worth about £20 at £20 an hour. This is a typical example, so swap in your own numbers.

    Want it done for you?

    We’re not IT people or a traditional agency. We’re business leaders who’ve put AI to work across sales, finance, operations and marketing. We can build your website using AI, write the words for you, train you to run it and hand it over, from £1,495 + VAT. Or we can keep it running for you with Website Care at £149 + VAT a month. See how our websites work, or look at our marketing service if it’s the content, not the site, that needs help.

    Growing pains like this are normal in a business that has grown quickly, and they’re fixable. Take the free Growing Pains Check to see where your business stands in three minutes, or book a free call about a Growing Pains Diagnostic: half a day with you and your business, then a written 90-day plan with the three changes that matter most.

    See how the workshops work →

    Or book a free 30-minute call with us →