Maciej Budzich, Business Process Automation Specialist
I find the manual work hiding in business processes and build automations that take it over.
I work in platform operations at Booksy, building automations around billing and subscriptions with Make, Google Apps Script, BigQuery and Claude. Where a mistake would cost real money, I keep a person's approval in the flow.
- Automated:BigQueryCollects the accounts
- Automated:AutomationChecks ambassador locations
- Automated:SlackCompares with the employee list
- Automated:EmailSends me the report
- A person decides:MeReview and apply every change
Selected work
Free subscription audit
A forgotten spreadsheet became a running check across 5 markets. The first full update removed about 50 employee accounts and 80 free subscriptions in the US alone.
Make, Google Apps Script, BigQuery, Slack
$3,000+ a month
In subscriptions that are no longer free, at the $30 minimum. The check runs by itself across 5 markets.
Billing fixes with Claude
About 100 accounts a week across 6 markets. Claude works through the cases from written instructions. My weekly time went from about 3 hours to about 20 minutes.
Claude, internal admin panel
3 h to 20 min
My time each week, by my estimate. About 500 accounts fixed so far.
Booksy Store seller settlements
A monthly settlement between Booksy, its sellers and accounting, rebuilt as one approval flow. In final testing, with full rollout expected at the December settlement.
Make, Google Apps Script, internal API, webhooks
In final testingBuilt end to end, from the internal API to accounting.
More automations
- Customer SMS alerts
- Two automations that replaced about 35 minutes of my daily manual work in an admin tool, both running for almost a year. One tells customers paying twice, through Apple and through Booksy, how to cancel one subscription. The other warns owners of possibly compromised accounts in the US and Spain, and messages every number recently tied to the account in case an attacker changed it. Together they send about 200 messages a month, at hours agreed with Customer Success.
- Mobile payment plan changes
- A SQL query finds accounts on the wrong mobile payment plan, and a daily automation changes the plan through the admin panel API. A few accounts a week, no manual work.
- AI ticket routingIn testing
- Claude in Make reads each new Jira ticket, compares it with 200 past tickets and picks the right teammate, who gets a Slack message. Between 40% and 70% match, it sends the ticket with a note that it's unsure. Below 40%, the ticket stays in the queue because it probably belongs to another team.
- Payment process analysis
- A Google Apps Script report on a process correcting wrong payments, used by my Team Lead and Head to judge whether it was worth running. It showed what customers paid, what they should have paid, and the monthly gain or loss, including customers who cancelled when asked to pay the correct amount. The best batches brought about $5,000 in profit.
- Automation dashboard
- A small local tool, built with AI, showing only my Make automations, their usage and errors, next to related Jira and Confluence items. It turns red when something fails, so I don't have to keep Make open.
How I build
Find out about failures first.
My dashboard turns red when a scenario fails, and repeat accounts in the weekly billing report get flagged to me.
Count the net effect.
My payment process report counted customers who cancelled, not just the money recovered, because the real question was whether the process paid off.
Keep a person on the expensive decisions.
Seller settlements need an employee's approval before anything goes out. The subscription audit reports accounts but never deletes them.
Decide what happens when nobody answers.
A process with people in it needs a defined outcome for silence. In seller settlements, the deadline decides.
Code background
I've been programming for about six years. It's why I can go beyond no-code tools when a process needs it.
AutoSQL: a system for a driving school
Built alone for a client in Poznań in 2025. Students get a website, a sign-up form and a calendar for booking lessons, and each booking is added to the instructor's Google Calendar automatically. The instructor gets a full admin panel: student details, calendar, adding and moving lessons, progress and hours driven, payment tracking, and first aid sessions with attendance.
Next.js, TypeScript, Tailwind, Supabase ·autosql.pl
Pokimoni
A non-commercial fan game I first wrote alone, then rebuilt with Claude Code. I owned the rules, game economy, acceptance testing and deployments, and the agent wrote the code.
Skills
| Group | Tools | Where |
|---|---|---|
| Automation | Make, Google Apps Script, webhooks, REST APIs | All automations |
| Data | SQL, Google BigQuery, Google Sheets | Subscription audit, SMS alerts, payment plan changes |
| AI | Claude | Billing fixes, ticket routing, Pokimoni |
| Programming | JavaScript, TypeScript, Python (core language) | JavaScript in Apps Script, TypeScript in AutoSQL |
| Web | Next.js, Tailwind, Supabase | AutoSQL |
| Work tools | Jira, Confluence, Slack, Salesforce | Daily work at Booksy |
About
I started programming about six years ago in technical school, just because I enjoyed it. I built websites and web apps for fun, along with a couple of commercial projects. My passion for automation comes from there: making something myself and seeing the real effect it has.
In 2024 I joined Booksy in operations. The role needed little more than spreadsheets, but I kept seeing repetitive work I could automate, so I started building on my own initiative. For over a year I've been largely on my own running billing operations for the US market, and in January 2026 I was promoted from Junior to Mid for that independence and technical work.
I study Computer Science part-time, and automation is what I want to do full-time.
Experience
- Mid Platform Operations Specialist, BooksyJan 2026 to present
- Junior US Operations Specialist, BooksyNov 2024 to Dec 2025
Education
- Computer Science, part-time, Poznan University of TechnologyOct 2024 to expected Jun 2028
Languages
- Polish (native), English (B2/B2+)