Automate email data entry. Keep people in control.
If someone on your team reads emails and types the details into another system, that job is worth automating. The skill is deciding what the system handles alone, how it proves each entry is right and what it hands back to a person. Here’s how I do it.
From an inbox to a checked record.
I built and maintain a workflow that turns school absence emails in Gmail into attendance records in eTap. It follows programmed rules, checks every saved entry and gives the attendance officer a morning handover.
The officer says it saves over an hour every school day, closer to two. It has been in daily use for about a month, as of September 2026.
Time saving is the attendance officer’s estimate, not a timed measurement.
Define a correct entry before building.
Start with the person who does the task. Work out where the information comes from, which record it belongs to and which fields must be saved. Plan for missing details, repeat messages and records someone has already updated.
For attendance, that means matching the student, checking dates and sessions, choosing the attendance code and keeping the parent’s comment. A doubtful match goes to a review queue. The system needs a clear line around what it can decide.
Connect the tools people already use.
First I check whether the existing software has an API, a connector or an import. Where browser automation is the right call, it needs checks around sign-in, saving and changes to the interface.
The attendance system uses the Gmail API and Playwright browser automation for eTap. It depends on an active session, the school computer and the vendor’s interface, so maintenance is part of the job. Before a prototype becomes something people rely on every day, it needs an owner.
Use the simplest approach that fits.
If the decisions follow clear criteria, start with programmed rules. Less structured information can justify AI, tested on real examples, with agreed data access and a clear route to human review.
No generative AI reads the attendance emails; the decisions follow programmed rules. In Just the Good News, AI selection and summaries are part of the publishing workflow. That choice shapes how you build and test the system.
A successful click is not a saved record.
Decide what proves the job is done. The attendance worker checks the saved code and comment before it marks an email as read. It also catches duplicates and protects existing codes.
For any other process, I agree those checks before anything runs unattended: what to verify, when to stop and how staff recover an unfinished task.
Make the remaining work easy to see.
At 8:45am the attendance system sends a summary of what’s done and what needs follow-up. Staff take over for the day, and anything uncertain or failed is waiting for them to review.
The handover is part of the product. Anyone should be able to see what happened, what’s left and whether the automation needs attention, without reading code.
Measure the work that actually disappears.
In a pilot I compare staff time before and after, including time spent checking exceptions and maintaining the system. I count verified completions, items needing review and corrections separately from emails received.
That gives you real numbers for deciding whether to extend it. Order requests, bookings and other email-driven admin often follow similar steps, each with its own rules and checks.
Explore my work in automation and custom softwareLet's build it.
Hiring for a senior creative technology role, or planning a project that matters? Tell me what you’re trying to achieve and where you’re based. I’m in New Zealand, spent six years in Singapore and am open to remote work or relocation.