If you read the post about using Claude to plan a mobile dog grooming startup, you saw a solid, structured business plan. The numbers were plausible. The advice was sensible. And that is precisely the problem. The hardest part of a startup is not the plan. It is the execution. Specifically, it is getting your first ten paying customers to believe you exist.
An AI cannot do that for you. It cannot make the first call, send the first email, or stand in a park handing out flyers while it rains. The objection is real: a plan is just theory. Revenue is reality. Let us talk about what happens after the plan.
The Gap Between a Plan and a Pipeline
Claude’s plan suggested marketing tactics: Google Business Profile, local Facebook groups, partnerships. This is correct. It is also the same list every blog post has written since 2015. The friction starts when you try to activate those channels.
Setting up a Google Business Profile takes an afternoon. Getting it to show up for relevant searches takes months of consistent signals—reviews, posts, citations—that you must generate. Posting in local Facebook groups requires you to navigate rules against self-promotion, build genuine engagement first, and hope the algorithm shows your post to anyone. Partnering with vets means you need a pitch, a follow-up sequence, and a way to track who you have spoken to, all while you are also learning how to groom a nervous cocker spaniel.
This is the real work. It is a sales and outreach problem disguised as a marketing task. And it is entirely manual, human, and slow.
What Getting the First Ten Customers Actually Looks Like
Forget broad marketing. For a solo founder with a service business, early traction is a direct, one-to-one sales campaign. Here is a more honest week-one plan after the business plan is done.
First, you build a list of 100 target customers in your chosen suburbs. You do not guess. You use a tool like Apollo or Clay to pull homeowners from specific postcodes, or you manually scour Facebook groups and Nextdoor for people who post about their dogs. You get names, where possible.
Second, you draft a short, plain-email introduction. Not a brochure. Three sentences: who you are, what you do, and a clear call to book a slot. You personalise each one with the neighbourhood name. You send 20 per day, manually, from your personal email. You follow up on day three and day seven if you hear nothing.
Third, you do the physical work. You print 200 double-sided flyers with a simple offer—£10 off first groom. You walk the streets of Didsbury and Chorlton for two hours every evening, placing them in letterboxes. You talk to every dog walker you see.
This is not in the AI plan. It is tedious, rejection-heavy, and absolutely critical. It is sales.
Where an AI Assistant Actually Helps (And Where It Doesn't)
This is not an argument against using AI. It is an argument for using it on the right layer of the problem.
Claude is excellent for structuring your thinking, as the original post showed. It is useful for drafting your service menu, writing a first pass of your website copy, or generating a checklist of legal requirements. It is a thinking partner for the planning phase.
But it cannot execute the outreach. It cannot decide which homeowner on your list looks most likely to have a high-maintenance dog. It cannot read a Facebook post and sense the frustration of someone struggling to book a groomer. It cannot adapt your flyer copy because you noticed people were asking about puppy-specific cuts.
The moment you need to move from theory to human conversation, the AI stops being the driver and becomes a tool for the driver. Your job is to steer.
Building a System That Outlives the Hustle
The goal of that brutal first-customer hustle is to build a system that eventually works without you doing every single task manually. But the system must be built on real data from real customers.
You use the first ten customers to learn. What did they actually ask about? What was their main objection? How did they find you? You feed that back into your messaging. You discover that 70% of your bookings come from Nextdoor, not Facebook. So you double down there. You learn that your £55 full-groom package rarely sells, but your £35 wash-and-tidy books out instantly. You adjust your pricing and marketing.
Then, and only then, can you start to scale the parts that are working. Maybe you use a tool like MiraReach to systematise your partnership outreach to vets, because now you know what message works. You set up a simple CRM to track leads from your flyer campaign. You create a repeatable process.
The AI-generated plan gave you the map. The sales work gives you the vehicle and the fuel.
What we would do next
If you are using Claude or any AI to plan your startup, treat its output as version 0.1. Then immediately shift your focus to the version 0.2 problem: how you will talk to your first 50 potential customers. Build that list. Draft that email. Walk that street.
The difference between a hobby and a business is not a plan. It is a purchase order. Go get one.
— Mira