Small Biz
AI Marketing

The Prompt Shop Podcast

The best brains in the room, and one of them better be yours.

Short, practical episodes on making better marketing decisions with AI, from a marketer who has run it inside real small businesses. Real client stories, the issues owners are grappling with now, and why quick wins today do not always win long term.

Explore first. Then plan. Then build something that lasts.

5 to 10 minutes. New episodes on Wednesdays. Hosted by Dan MacInnis.
Listen now Apple Podcasts Spotify Podomatic

Latest episode

Episode 19. One job, start to finish: how this podcast gets made

Episodes and show notes

Start at Ep 1, or pick a topic. New episodes on Wednesdays.

Ep 1 Explore first: the weekend that changed how I work Why the opportunity comes before the plan, and how to spend your first weekend with AI. 5 min 41 sec

What this episode covers

  • Why Dan is back in the studio after 11 years of podcasting: small businesses are grappling with another big change, and this time it is AI.
  • Easter 2025: four days playing with ChatGPT, from planning a holiday and tidying drafts, to realising it could be customised to work like another employee in the business.
  • The tools will not replace you. They change how you think: fewer low-level tasks, more room for strategic decisions.
  • The trap of AI slop: more content and more output is not the point. Better decisions are.
  • Without a harness on the tool, and without customising it, it will not serve your business as well as it could.
  • Step one: spend a weekend just playing. ChatGPT, Claude, Copilot, or all three. Ask it questions. Don't buy anything, don't build anything.
  • Step two: customise it. Find the personalisation area in whichever tool you use (ask the tool if you can't find it), then ask it to interview you about your business, how you work, your ideal customer and what makes your brand different.
  • Why that matters: an uncustomised tool gives you flatpack advice, the same as everyone else gets. Customised, it becomes custom made for your business. Output improves tenfold.
  • Next episode: how to build skills and workflows that streamline how you work.

Resources mentioned

Read the transcript

I'm Dan MacInnis, and I did a podcast for about 11 years, almost 15 years ago. But I've jumped back into the recording studio because what I've found is small businesses are now grappling with another change, and that is AI. So this podcast is about how you leverage AI for small business marketing, what you need to know, and how you get the best brains in the room, making sure that one of them is yours.

I started with AI in April last year. It was over the Easter break, and I was really just playing with ChatGPT. I was doing the usual things: planning my next holiday, asking it questions, researching, taking a draft of something I'd written and trying to make it sound better. Over those four days I went from understanding ChatGPT as a resource of knowledge to understanding how to customise it to make it like another employee for my business.

I imagine a lot of you are playing with lots of tools and trying to work out how this could be a point of leverage for your business. That's what I encourage you to do. Take what you do in your business and see how these tools can improve it. They can't replace you. It's a new way of thinking: using the tools to improve your decision-making and take away the low-level tasks, so you can do more high-level strategic thinking.

It's about thinking about how you work with AI, not just creating more of what they call AI slop, which is just more content, more output. I know it's tempting, but this is a really powerful shift in the way people work, particularly for small businesses. It gives you the ability to be an expert in so many areas because it can take over tasks you're doing manually or outsourcing. But without a harness on whichever tool you're using, and without customising it, it isn't going to serve your business as well as it could.

So let's talk about the first thing I would do. Spend a weekend just playing with it if you've never used ChatGPT, Claude or Copilot before. Perhaps use all of them. Just ask it questions. Just play. After you've done that for one weekend, the next thing is to customise it so it's useful for you.

What does that mean? In each tool there's an area of personalisation. Whenever you're in doubt, just ask the tool and it'll show you where it is. Ask it to interview you so you can customise it for your business. It'll ask what your business is about, how you work, who your ideal customer is, what makes your brand different. If you upload that into the personalised section of the tool, whichever tool it is, the output improves tenfold. Instead of general advice it starts to fit you. It recommends things that are much more relevant. That gives you a little insight into how powerful this platform is, and then you can start to harness it for much better workflows and outputs.

So first play, then customise. Next episode we'll talk about how to build skills and workflows that will streamline how you work.

Ep 2 Teach it your customers, your competitors and how you decide Build the context bank that turns generic AI answers into ones that fit your business. 5 min 43 sec

What this episode covers

  • Your AI only knows what you have told it. Last episode was play, then get the tool to interview you. This one goes deeper.
  • Teach ChatGPT, Claude or Copilot who your customers actually are, what they worry about and why they would choose you over a competitor.
  • Then ask it to read its understanding back to you, and correct what it gets wrong. That correction is the work.
  • A quick test to see whether AI already recommends your business.
  • Why FAQs written as real questions help you get found by AI.
  • Feeding competitor research back into your project folder.
  • The customisation most people skip: telling the tool how you like to make decisions.
  • Next episode: memory and skills.

Resources mentioned

Ep 3 Memory and skills, the shared brain and the tasks it does on repeat Why memory matters, what a skill actually is, and the three questions to ask before you install one. 6 min 23 sec

What this episode covers

  • Two settings turn a chat window into a coworker: memory, and skills.
  • Everything you have fed the tool over the last two episodes builds a shared brain between you and it, as long as memory is switched on, whichever tool you use.
  • Skills: specialised, repeatable tasks you create or download so the tool does the job the same way every time.
  • Where skills live, and what a GitHub repo is in plain terms.
  • Why every skill you add uses part of your setup, and why a lean list beats a loaded one.
  • The three questions to ask your AI before you install a skill.
  • Next episode: agents, groups of skills in a workflow, and how work gets done.

Resources mentioned

Ep 4 How agents get your work done If it has a trigger, inputs and steps in an order, and it runs without you, it is an agent. 5 min 56 sec

What this episode covers

  • What an agent is, in one sentence, and how it differs from a skill.
  • The 7am agent: reads Drive, Gmail and the calendar, flags what is urgent and drafts the replies. The first version was email triage; it became a priorities dashboard.
  • The monthly marketing report as an agent: pulls from analytics, your email platform and social, and lands as a dashboard or a deck.
  • The easiest one to try first: a clean-up task that filters promotional email out of Gmail or Outlook every 30 days.
  • Why you need a paid plan for these to run, and why the first version usually is not the one you keep.
  • The only test that matters: are you using what it gives you, or working around it?
  • Next episode: the knowledge base, how to give your AI more to work with than its memory can hold.

Resources mentioned

Ep 5 The knowledge base that stops you pasting the same documents into every chat Point it at your filing cabinet. Stop re-explaining your business every time you open a chat. 9 min 59 sec

What this episode covers

  • The knowledge base is the filing cabinet for your business: sales, marketing, admin, HR, as deep as you want to go, and Claude or ChatGPT pulls from it whenever it works for you.
  • Dan uses Notion. It handles databases, connects natively to Claude and ChatGPT, and is not expensive. Dropbox, Drive or Slack can work, but keep it all in one repository.
  • The test: how many times have you pasted the same document into the window? More than twice and you need this.
  • Memory is what it remembers about you. The knowledge base is what it looks up about your business. It beats memory for anything you would otherwise keep in your drive.
  • Ask the AI to build the scaffolding first, then upload into it. Convert documents to Markdown (.md) so they cost fewer tokens.
  • File a good proposal and a poor one side by side, and your monthly reports, then set a task to analyse the last four months.
  • Your style guide goes in first: fonts, colours, logo, tone of voice, value proposition. Then build a design system in paid Claude and apply it to every output.
  • Next episode: Connect it, plugging Claude into your email, calendar and drive.

Resources mentioned

Ep 6 Connect it: stop being the courier between your inbox and your AI Plug Claude or ChatGPT into your email, calendar and drive, in the right order, and get out of the middle. 5 min 4 sec

What this episode covers

  • Until your AI is connected to your tools, you are the courier: you carry the email in, do the draft, paste the calendar into the chat.
  • Where to look: Settings, then Connectors.
  • The two connections to make first: the email you send from and the calendar you live in. Triage, meeting prep, invites and notes are what they unlock for a service business.
  • Your drive is next. Banking is never.
  • Read the permission levels before you switch anything on. Start with read access.
  • The safety habit: draft before anything sends, deletes or books.
  • Next episode: repeated tasks, the Monday dashboard.

Resources mentioned

Ep 7 Your Monday dashboard One scheduled task that tells you what matters this week, and why. 5 min 44 sec

What this episode covers

  • Once memory, skills, connections and the knowledge base are in place, the next step is a dashboard: your priorities, in one report, every Monday.
  • A dashboard is a report, not software. It pops up in your AI and gives you guidance.
  • Dan's reads Gmail, her calendar, what went out on social and what is due from the knowledge base, then prioritises what matters and why.
  • Start small: a few items, a recap of what is due in the next few days. Yours might be schedule, sales or customer service.
  • Add a narrative: ask Claude or ChatGPT to spar with you on how you decide what is most important.
  • Fix the highest-value things first. Admin that comes up time and again is the signal to give it its own scheduled task.
  • Action this week: set it up with Gmail and your calendar only. Two sections.
  • Next episode: which one for which job, ChatGPT or Claude.

Resources mentioned

Ep 8 Which one for which job. ChatGPT or Claude? The wrong question, and the card Dan uses instead. 6 min 21 sec

What this episode covers

  • Dan started on ChatGPT and found it easy, good for copy and fine for tasks, then hit the wall when she wanted agents and workflows. Claude scaled to her connections.
  • The honest answer: the differences are now minimal. You can export from either and move. Nothing is locked in.
  • She still uses both: ChatGPT for research and a second angle on copy, Claude for the infrastructure her business runs on. One paid, one free.
  • The rule is not which tool. It is one tool holds your context, and you use the other to check it.
  • The real change is using one as a discipline and thinking beyond tasks to the whole picture.
  • Next episode: Claude as a real sparring partner, the rules that keep it honest.

Resources mentioned

Ep 9 Rules that keep it honest How to stop your AI agreeing with you. 6 min 19 sec

What this episode covers

  • Every AI tool is a yes machine. It is trained to be helpful and agreeable, so it tells you your plan is great, your headline is strong and your pricing makes sense.
  • Dangerous for a small business owner, because most of us only have ourselves to rely on and never get the pushback.
  • The fix is not a clever prompt. It is a short set of rules you write once into your customisation, so every response gets argued with.
  • Rule 1: respond like a senior expert in my niche, with examples of people whose work I aspire to.
  • Rule 2: be my sparring partner. Find my weak spots and blind spots. Do not default to agreeing with me.
  • Rule 3: be blunt. Tell me why, give me an option, explain your point of view.
  • Rule 4: before you run a prompt, ask me the sharp questions: audience, proof, sources, the outcome I want.
  • Rule 5: check the facts of every claim and reference it. Rule 6: challenge everything before executing it.
  • The second-tool check: paste the output into ChatGPT, Claude or Gemini and see if the answer changes, then iterate again.
  • Action this week: two rules into your customisation, then ask it about something you have already decided.

Resources mentioned

Ep 10 Will AI make my business feel less personal? The four rules that keep your business sounding like you. 6 min 23 sec

What this episode covers

  • The fear is not "will it take my job". It is "will it make my business feel like everybody else's". A US survey put that fear at two thirds of owners.
  • "Use AI less" is not a plan. The rule is simpler: AI does the work nobody feels, you do the work somebody feels.
  • Where personal actually lives: the reply, the apology, the visit or the job itself, and the choosing moment on your about page and your first quote.
  • Rule 1: the customer facing sentence is yours. Rewrite the first and last line of every AI draft by hand.
  • Rule 2: the apology never comes from a template.
  • Rule 3: teach it your voice before it writes for you. One page, pasted into your customisation.
  • Rule 4: say when it is AI, where it matters. A chatbot at 11pm says it is a chatbot.
  • The quarter of owners who lost business to customers using AI instead: if a chat window could replace you, you were selling information. Judgement and showing up are the business.
  • Some jobs still go to AI, and that is part of being efficient. Decide which touchpoints keep your human touch, especially where customers need your advice.
  • Action this week: take one AI drafted reply, rewrite it by hand, and know the difference, so you can train your staff on it.

Resources mentioned

Ep 11 What I still do not hand over to AI Three lines: what it prepares, what you approve, what stays your call. 8 min 23 sec

What this episode covers

  • Season two is about decisions: what you hand over, what you check, what stays with you
  • Before any email campaign, a person confirms who it is going to and how many. AI can write it; it does not approve the audience
  • Three lines for every job: hand over the preparation, hand over with a check, keep the decision. Three parts of the same job, not three kinds of business
  • Line one, preparation: the Monday dashboard, meeting notes, first drafts. It gathers and organises; you keep the source close enough to check anything you rely on
  • Line two, approval: give it the brief (customer, offer, reason, action) and it drafts. Before it becomes a campaign, check the offer, the claims, the links, the tone, and confirm the audience and the contact count. Scope and price always stay with Dan
  • It is only as good as what you set up in season one: tone of voice, ideal customer, products and services. No knowledge hub, generic output
  • Line three, the decision: should we send this at all, is the call to action clear, is the timing right, should some people get a call instead. A polished draft does not answer those
  • It only knows the context you gave it. It does not know about last Tuesday's phone call. Same for an apology or a complaint: you sit down and feed it everything, or it misrepresents you
  • Dan's boundary: nothing connected to banking or accounts. Only the specialist connections the business needs, such as email and a few software tools
  • A task done right earns more freedom. An action you cannot undo, like a change in your CRM, needs a human check. Set "don't allow until I've checked", never "always allow", so you can fix it when something goes wrong
  • Action this week: take three jobs you have had AI help with and write down what needs approval and what decision still stays with you
  • Humanise first, automate second

Resources mentioned

Ep 12 Should I take this client? The interview you skip when you want the work. 8 min 17 sec

What this episode covers

  • When you want the work, you stop asking questions. AI does not get hungry, so it runs the interview you skip
  • Set up once in a project: your capacity this quarter, your no-go rules, your customer profile
  • The instruction: interview me, one question at a time, push on anything that sounds like hope, only then give a go or no-go with a reason
  • The ten questions, from "what just happened that made them enquire this week" to "is this a client or a rescue" and "am I saying yes to feel safe"
  • The demo: a real enquiry from this year, names and sector swapped. An IT company that had lost a big account and wanted leads fast, two agencies in the last 18 months. The interview called it a rescue, not a client, and asked whether the budget was for a rescue
  • Set it up as a skill: "I have a new enquiry, run me through the interview." Give it your business goals separately so it knows what great looks like, not just the rules
  • You can overrule it or go and find out more. Either way it makes you take an honest look at fit
  • What it is not: the decision. It does not know what you promised on the phone or how much you need the cash. Interview me, not decide for me
  • Takeaway: use the Go or No-Go Interview sheet before you quote and before you say yes. An enquiry is not a fit just because it arrived
  • Humanise first, automate second

Resources mentioned

Ep 13 Is my price right? The number nobody has argued with, and how to make AI argue with it. 7 min 1 sec

What this episode covers

  • Most prices were set once, years ago, by looking at the competitor down the road and knocking ten percent off. Nobody has argued with the number since
  • AI cannot set your price. It can argue with the one you have, using the numbers you put in front of it, and it does not care if the answer is uncomfortable
  • Not a pricing masterclass, not value-based pricing theory, not "charge what you are worth". One decision: is the number I charge for this service or product right, and how do I know
  • Same pattern as Eps 11 and 12: set up a sparring partner once and it holds you to account
  • The pricing project holds three things. The offer in plain words: what is included, what is not, how long it takes, what it costs to deliver (for a product, what it costs to create, promote and ship). The last twelve months: quoted, won, pushed back on price, and what you charged if it varied. And three competitors' public prices or ranges, pasted in with the date
  • The instruction: spar with me on this price, do not give me a number first, ask what the outcome is worth to the customer, what happens if I raise it ten percent, and what I am leaving on the table. Argue higher, argue lower, using my numbers and the competitor prices. One recommendation and one thing that would change your mind
  • The test: when did someone last push back on your price? If never, it is too low. If everyone, you are attracting the wrong buyer
  • Reset the packaging, not just the price. Maybe it should not be an hourly rate. Maybe it should be a package that represents a value
  • What AI cannot see: your cash position, the client who always pays late, the offer that is a door-opener to further work. Those can change your call. It spars with what you tell it
  • Action this week: run the spar on your real offer, in whatever AI you use, using the sheet in the notes
  • Questions: leave a message at smallbizpromptshop.com.au or a comment wherever you listen
  • Humanise first, automate second

Resources mentioned

Ep 14 Which of my jobs actually go How AI is changing the shape of a small business, the tasks that go, the ones that change, and what the owner's job turns into. 7 min 30 sec

What this episode covers

  • The Australian data: AI augments more than it automates. Around 4% of jobs highly exposed, about a fifth medium to high, clerical and admin most exposed. Among businesses using AI, employment rose for nearly one in five and fell for about one in twenty
  • Every role, including yours, is a bundle of tasks: gathering and organising, judgement, showing up. AI takes the first, changes the second, barely touches the third
  • The question is which tasks inside each job go, and what the person does with the time
  • Set up the same way as Eps 11 to 13: every recurring task by role with hours, your goals for the year, one instruction. Sort into goes, changes, stays, say why, and show what each role becomes. Do not tell me to hire or fire
  • Three calls: the Monday report goes (the hour compiling it comes back), the client proposal changes (it drafts from your good example, you rewrite and check the numbers), the difficult conversation stays (the complaint, the price increase)
  • The owner's job turns from doing to checking to deciding. Season one built the system, season two is the decisions
  • Two warnings: clerical roles change first, sort their column with them not to them; and the businesses that grew used AI to do more, not the same work with fewer people. Freed hours need a destination
  • Action this week: list your own recurring tasks and hours, just yours, sort them into the three columns without AI first, then run the instruction and see where you disagree
  • Questions: smallbizpromptshop.com.au or a comment wherever you listen
  • Humanise first, automate second

Resources mentioned

Ep 15 Is my data safe in there? What goes into a chat, what never does, and the three questions to ask any tool before you connect it. 6 min 58 sec

What this episode covers

  • The Privacy Act applies the moment personal information goes into an AI tool. The OAIC's stated best practice: do not enter personal, especially sensitive, information into publicly available generative AI tools
  • Check whether you are exposed by memory or sharing settings. In ChatGPT the training setting sits under Data Controls, in Claude under Privacy; Copilot lives inside the Microsoft ecosystem. Look at yours today, because the default may not suit you. Business plans and tools used through other software are generally not used for training by default. Switching training off is not deletion: your words still went to someone else's computer
  • Line 1, goes in freely: your services, prices, tone of voice, drafts, your own notes. The season one knowledge base
  • Line 2, goes in with names removed: the enquiry, the complaint, the proposal. Swap the name and the identifying detail. Every demo on this show was done that way
  • Line 3, never goes in: anyone's health information, bank and tax file and card numbers, ID documents, passwords, anything given in confidence, banking and accounting connections
  • Three questions before connecting any tool: what can it read and do, is my data used to train it and where is that setting, what happens when I disconnect
  • Habit one: temporary or incognito chat for the one-off, still with names out. Habit two: a house rule staff can repeat, because four in ten Australian businesses found staff using unapproved AI tools last year
  • Action this week: open settings in your tool and find the training toggle, then read your last ten chats for anything that breaks the data line. A risk with your own information is one thing; a risk with a client's is another
  • Humanise first, automate second

Resources mentioned

Ep 16 What should I stop selling? What AI makes cheap, what to repackage around judgement, and the higher-value work the hours should go to. 7 min 19 sec

What this episode covers

  • If a chat window could replace what you sell, you were selling information (Ep 10). This episode is what you do about it
  • Australian evidence: businesses using AI products growing about 2.8 times faster (MYOB, April 2026); Treasury calls uptake widespread but shallow. Saving time is easy. Growing means doing different work
  • Set up the sparring project: twelve months of invoices by offer with names off, what each offer involves, your customer profile. Instruction: which parts could a customer now get for free, which could they not, sort into three columns, argue with me
  • Column one, stop or shrink: the generic content package, templates and checklists you charged for, the report that summarises numbers the client already has, first drafts sold as the product
  • Column two, repackage: this is where most of your hours should go. Take the offer where the information was the packaging and the judgement was the product, and sell the judgement. Sell what the client cannot get unless they talk to you
  • Column three, sell instead: the audit you never had time for, the implementation you used to hand off, the strategy day that was always next quarter
  • What it cannot see: a low-value offer can still open the door to good clients. You do not have to cut everything, but you need a legitimate reason for keeping it
  • Where the podcast goes next: the first twenty episodes are foundations; from episode 20, real businesses where Dan has put this in and what it did to the numbers
  • Action this week: top three offers by revenue, and for each write what a customer could now get for free and what they cannot. The second list is what you repackage first

Resources mentioned

Ep 17 The cheat sheet Ten quick wins, five things to watch, and the five questions small business owners still ask. 8 min 11 sec

What this episode covers

  • Ten quick wins: first and last line by hand; paste the enquiry, names off; file a good and a bad proposal and point it at both; ask for the five questions before the answer; paste the draft into another tool and ask what is weak (Dan runs Claude to ChatGPT to Perplexity to Grok to Gemini and combines them); record two minutes and repurpose from the transcript; one sentence of context beats ten instructions; Monday dashboard with inbox and calendar only; the never list (delighted, seamless, elevate, reach out, unlock) into your customisation first; go or no-go before the quote
  • Five things to watch: it agrees with you unless you wrote the rule (Ep 9); a wrong number sounds like a right one, so figures, dates and legal points get a source; never "always allow" on an action you cannot undo (Ep 11); names out of the chat, health, bank and ID never (Ep 15); saving time is not growing (Ep 16)
  • Which tool? The one that holds your context (Ep 8). Paid plan? For the tool that holds your business, yes; free for the second. Client work safe? Names removed yes, health, bank, ID never, and check the training setting. Sound like everyone else? Only if it writes before you teach it your voice (Ep 10). Where to start? Ep 1, then the knowledge base, then connect the inbox, one a week
  • The takeaway is the cheat sheet, one page, stuck where you have your coffee
  • Questions and comments welcome; Dan answers them in the next episode

Resources mentioned

Ep 18 Will AI recommend my business? What ChatGPT and Perplexity actually read, why your about page matters more than your homepage, and the first three fixes. 8 min 43 sec

What this episode covers

  • The one-minute test: ask ChatGPT, Perplexity or Claude to recommend a business like yours in your area, and look for your name
  • Dan's own audit earlier this year: 53 out of 100 for AI search health. Homepage with no headings, not one, and the same on the shop and contact pages. Tags and categories corrected across every blog. Twelve posts on gibberish URLs. The FAQ page rewritten, because people ask AI questions and a site that answers them gets recommended
  • When you write a page or a blog, ask your AI to check it against SEO best practice before it goes up
  • What AI reads on your site: an answer in the first paragraph, headings that say what the page is about, specifics a machine can lift, one consistent business entity, crawlers allowed in
  • What AI reads elsewhere: a US study by Fractl found brands with strong search rankings could be nearly invisible to AI, and the difference was third-party coverage. What others say about you matters more than what you say. Large brands, not Australian small business, but the logic transfers
  • The wrong-box problem: a bookkeeper who mostly does trades gets filed under accounting and never surfaces for "bookkeeper for tradies". Your category is a decision, and it matters most in directories and Google Business Profile
  • Fix one: the about page, facts not hype. "More than sixty Geelong trades businesses since 2014" beats "passionate about exceptional solutions". Same words everywhere
  • Fix two: structure. A heading on every page, the answer in the first paragraph, specifics in the body. Then the housekeeping: kill thin archive pages, fix bad URLs, set up redirects for old pages, read your FAQ as if you were a bot. A weekend job
  • Fix three: be mentioned somewhere that is not your site. Association member list, local directory with a real description, a supplier or partner page. Authority as well as visibility
  • Action this week: run the one-minute test, see where you are, and set about improving it

Resources mentioned

Ep 19 One job, start to finish: how this podcast gets made Twelve steps, one workflow, about forty minutes of Dan's time. 4 min 58 sec

What this episode covers

  • The last of the foundations, and the promised whole-process episode. The best example is this podcast: a workflow built with AI over five weeks using everything from Eps 1 to 18
  • The job: twelve things every episode needs, from the recording and the mix to the transcript, show notes, Podomatic description, takeaway sheet, page, subscriber email and Substack essay. Hours of work, now about forty minutes of Dan's time
  • Preparation: a Friday hand-over. The schedule of what gets recorded comes to Dan by email, she builds out the episode, then records
  • Recording: handed to Claude. It adds the intro and outro, checks the levels, mixes it and sends back the final
  • Upload: Dan puts the final into Podomatic, and it streams to Apple, Spotify and the rest automatically
  • Compared with running this podcast eleven years ago, most of the admin is gone and the content and takeaways are more professional
  • The method: connect the task step by step into Claude, then connect the tools it needs. Systematic, and it cuts the admin
  • What it cost: time to set up the first time and a few test runs. Then it runs on rules: memory, skills, a knowledge base, brand rules, and connections to Drive and the email platform
  • Action this week: pick one recurring job with more than six steps, write each step down, then ask Claude how to automate it. The Workflow Map in the notes helps

Resources mentioned

Ep 20 The practice manager who stopped paying in the dark Coming soon. The first real business: a Melbourne physio clinic, and the system its practice manager now runs herself.

From here on, real businesses and real numbers. A practice manager who "didn't know marketing" went from paying for SEO and a website she could not see, to rebuilding the site in Claude herself, connecting the tools into one system, and putting the time back into the clinic. It is booked out.

Done the weekend? Find your stage.

Two minutes, no email required to see the result. Take the quiz, then tell Dan what surprised you.

Take the quiz