Replacing HoneyBook with a tool that fits
I was paying every month for an all-in-one tool that did about 20% of my job and 100% of someone else's. So I built exactly what the business needed and stopped renting the rest. The payoff wasn't the money — it was the evenings back.
- 1 subscriptionreplaced outright, not supplemented
- 4 jobsrunning quietly inside the existing workflow
- 1 screenfor the owner's view, instead of five
- Client
- Coffee Cart Boys — mobile coffee catering
- Role
- Owner, and the person who built it
- Timeline
- Built in the gaps around running the business
- Built with
- Next.js · Postgres · Prisma · Claude
The problem
Coffee Cart Boys is a mobile coffee catering business. The work is events: an inquiry comes in, it needs a quote, the quote needs chasing, and then the job needs to actually happen on a Saturday morning with the right gear in the van.
The software situation was the ordinary small-business one. An all-in-one platform handled bookings and invoices in a shape someone else had decided on, a spreadsheet handled the parts it didn't, and my inbox handled everything that fell between the two. None of it was broken. It was just a monthly bill for a tool doing about a fifth of the job, plus a lot of unpaid admin filling in the rest.
The tell was where my evenings went. Not to the coffee — to re-typing an inquiry into a spreadsheet, to rebuilding the same quote from a blank page, to remembering three days late that a lead had gone quiet.
The approach
I didn't start with a feature list. I started by writing down what I actually did in a week and marking the parts a computer should have been doing. Four things came out of that: intake, quotes, follow-up, and the owner's view.
Then each one got built as a small job inside the workflow that already existed — not a new place to go, not a new habit to learn. The rule I held myself to: if using it required me to remember something, it wasn't finished.
The AI in it is deliberately unremarkable. It reads a messy inbound message and produces a tidy record. It drafts a quote from real pricing that I read and adjust before anything is sent. There is no chatbot. There is no screen that says "AI" on it. Every generated thing passes under my eyes before a client sees it, because the judgment is the part of the job I'm actually being paid for.
What it changed
The subscription is gone — replaced, not supplemented, which is the only version of that sentence that counts. But the real difference is the follow-up job, which recovered bookings I used to lose purely to my own inbox. That's revenue that existed the whole time and was falling through a crack in my attention.
The owner's dashboard did something subtler. Knowing where the business stands used to require assembling it in my head from five places. Now it's one screen, which means I check it more, which means I catch things earlier.
Why this is the first case study
It's the honest one. I didn't build this to sell a service — I built it because I was the customer, and then found that the same shape of problem sits in most small businesses I talk to. A tool that does a fraction of your job, a person filling in the rest by hand, and no one ever quite having the time to fix it.
If it can replace the software I was paying for, it can replace the software you're paying for. That's the whole idea.
The jobs it quietly runs.
Intake
An inquiry — form, email, or text — lands as a clean client record automatically. The messy version in, the tidy version out. No re-typing into a spreadsheet.
Quotes
From that record, a drafted quote is ready in one click, using real pricing and the details of the request. I read it, adjust, and send. No blank page.
Follow-up
If a lead goes quiet, a personal, well-timed nudge goes out on a schedule I set — the note I'd have sent if I'd remembered. This alone recovered lost bookings.
Owner's dashboard
Everything scattered across tools — who's booked, what's outstanding, what's coming up — pulled into one at-a-glance screen. One page instead of five.
None of this is flashy. Nothing to show off. The AI does small jobs inside the workflow I already had — and the result is a week with less admin in it.
Want the full teardown, field by field? Read the write-up →