Your team already uses AI, in eight personal ways, on accounts you can’t see. One plugin packages one process so everyone runs it your way. A menu of them turns your business into an asset that works without you in the room.
Now the only question left is the practical one.
How do you actually get from here to there?
I have a thesis on this as well:
The businesses that get this right don’t start trying to get mass adoption. They start with one process, get it live, and let the win do the talking.
The ones that fail try to package everything at once and drown.
So let’s do it properly.
Start With One
I know what the temptation is. You’ve read three posts. You’ve spotted six processes in your business that need packaging. You want the whole menu by Christmas.
Don’t.
I’ve watched this movie in the corporate world for twenty years. Big system, big launch, everyone trained on everything. Six months later, nobody uses any of it, and the project gets quietly buried.
One plugin. One team. One clear win.
Get it live. Let the office see the quote plugin build a proper quote in two minutes. Let the estimator see his rules running without him. Then they’ll ask you what’s next, and you won’t have to sell the second one at all.
#1 – How To Pick The First One
Not every process makes a good first plugin. Here’s the filter I use with every client.
Stuck In One Head
The process only one person really knows. The estimator’s spreadsheet. The office manager’s way of chasing suppliers. If that person going on holiday causes a problem, you’ve found a candidate.
Done Often
Weekly at minimum. Ideally daily. A plugin for something you do twice a year proves nothing. A plugin for something you do twenty times a week proves itself in a fortnight.
Done Inconsistently
The process where quality depends on who picked it up. If two people would produce two different quotes from the same job details, the plugin will show its value straight away.
You Can Describe It
Can you explain the process in a few sentences? “Quotes get built in three sections, labour comes from the rates sheet, we never quote without a site visit.”
That’s a first plugin.
If explaining it takes an hour and fourteen exceptions, save it for plugin number four.
For most service businesses, quoting comes out top. High frequency, money attached, and almost always inconsistent. With the construction client, it was the tender process. Same thing, bigger documents.
#2 – The Build, Stage By Stage
Here’s exactly what happens once you’ve picked one.
Map (Week 1)
One call. Online, an hour or so.
My job on this call is to get the process out of your head, or out of your best person’s head, and onto paper.
This is the same conversation I’ve had hundreds of times as a developer. Sit with the person who knows the process. Ask how they do it. Then ask about the exceptions, because the exceptions are where everything falls over.
“So you always send the quote within 48 hours?”
“Yeah. Unless it’s over twenty grand, then Dave looks at it first.”
That “unless” is the whole job. We find every one of them.
You bring real examples. Three or four good quotes you’ve sent. A bad one, if you’re brave. That’s how the plugin learns your standard and your voice.
What you do: one call, dig out some examples.
What I do: turn it into a written specification you can read and sign off.
Package (Weeks 1 to 2)
I build the plugin.
Your rules become the skills, the one-liners become the commands. Your tools get connected through the proper routes the software companies built for this.
And I bake in the guardrails. Anything that leaves the business, a human sees first. Anything the plugin isn’t sure about, it stops and asks. That’s not optional. I won’t build it any other way.
What you do: approve a couple of access permissions. Same “allow this app” screen you’ve clicked a dozen times.
What I do: everything else.
Pilot (Weeks 2 to 3)
One or two people use it on real work.
Not the whole team. Not yet. Your best person and someone newer. Because if the newer one gets the same result as your best, you’ve proved the point.
Every output gets reviewed. The “that’s not quite how we’d say it” gets fed back and fixed. Every exception we missed on the call turns up here, and we add it.
With the construction client, this stage found things nobody mentioned on the map call. A supplier who always gets a different drawing set. A trade that needed splitting two ways. Normal. That’s what the pilot is for.
What you do: use it, and tell me where it’s wrong.
What I do: fix it, fast.
Roll Out (Week 3)
The whole team gets it. Thirty minutes of showing them where it lives and how to run it. If they can use an app on their phone, they can use this.
Then the important bit. It becomes the default. Not “you can use this if you want”. This is how we build quotes now. The old way goes in the drawer.
What you do: back it, out loud, in front of the team.
What I do: be around for the first week of questions.
Keep It Current (Ongoing)
The stage everyone forgets to plan for.
Your tools update. Claude updates, often. Your process changes because your business changes. A plugin nobody looks after drifts, and it drifts quietly. Nobody gets an error. The quotes just slowly stop matching how you actually work now, and you find out when a client points it out.
So every plugin I put live stays looked after. Tested after updates. Updated when your tools move. Changed when your process changes. You get a monthly note showing what ran, what changed, and anything that needs a decision from you.
What you do: read the note.
What I do: everything in it.
#3 – Timeline And Cost
Straight numbers. You’ve earned them by Part 4.
Timeline
Map, week one. Package, weeks one to two. Pilot, weeks two to three. Roll out, week three.
From first call to live plugin: two to four weeks. Most of that is my time, not yours. Your total commitment is one call, a fortnight of using it and telling me what’s wrong, and thirty minutes with the team.
The Build
A first plugin, mapped, packaged, piloted and rolled out, sits in my Foundation tier. £3,000 to £5,000, depending on how many tools it connects to and how many “unless” moments your process has. Half up front, half when it’s live.
Building The Menu
Plugin two is faster and usually cheaper than plugin one. Your tools are already connected. Your team already knows the drill. Each new one is another £150 a month, and another process that stops living in someone’s head.
Set that against what it costs to hire someone who knows all your processes, and then train them, and then hope they stay. This isn’t a close decision.
#4 – What You’re Actually Buying
Strip the technology out and here’s the product.
A new starter builds a quote on day one that looks like your best person built it.
The estimator goes to Tenerife and tenders still get priced.
The complaint that lands at four on a Friday gets handled your way, without you.
And the 2 am question, “would this place survive without me for a month”, gets a better answer every time you add a plugin to the menu.
You’re not buying software. You’re buying a business that runs to your standard when you’re not looking at it.
Conclusion: Getting Your First Plugin Built
Going back to what we covered:
- Start with one, and let the win sell the second
- Pick the process stuck in one head, done often, done inconsistently
- Map, package, pilot, roll out, keep it current
- Two to four weeks, £3,000 to £5,000
That’s the whole series. Your team already uses AI badly. A plugin packages your process. A marketplace packages your business. And your first one is three weeks away.
The technology isn’t the hard part. Anthropic built the platform. I build the plugins.
The hard part is deciding your business is worth writing down. Most owners never get round to it, because there’s always a quote to get out.
You’ve read four posts. I’d say you’ve got round to it.
I hope this helps.
Oh, just one more thing. If you want to talk about which process goes first in your business, book a call. Bring your messiest one. I’ll tell you honestly whether it’s the right place to start, and if it’s not, which one is.
