GuidesHow to2 min
How to replace a spreadsheet with software you own
Map the real columns, agree who confirms a row, propose-then-confirm, then the sheet becomes an export — not the company.
If two people cannot edit the same Google Sheet without breaking a job, you do not need a keynote on agents. You need a queue. Map the real columns, agree who confirms a row, propose then confirm. After go-live the sheet is an export — not the company.
This is core Invent work at SINLE: a system of record you own, on a domain you keep, with a human lock if a clerk (human or model) drafts the row. AI is optional on top of a table. The table is not optional. You own the repo on final payment.
The sequence below is the migration. Fire the new path on a Friday as a test. If Monday cannot run, you bought a boss. Undo it.
What changes on go-live day
The sheet’s real columns become a table: customer, job, due date, amount, status, owner. A clerk may draft a row. A named person confirms it against the original file. After that, the sheet is a nightly export. Nobody types into it as the source of truth. If Monday cannot run without the sheet, you undo the cutover. You bought a boss.
AI on top is optional. The table is not. You own the repo on final payment.
The sequence
Do these in order. Skip a line and you are back to a workshop.
- 01
Export the sheet and name the real columns
Customer, job, due date, amount, status, owner. Delete zombie columns.
- 02
Agree who may confirm a row
If two people can silently overwrite, you will not trust a queue either.
- 03
Stand up a mailbox you control
PDFs and emails land there. Nothing goes to a consumer chatbot.
- 04
Propose, then confirm
A clerk — human or model — types the draft row. A person sees the original file and hits confirm.
- 05
Write confirmed rows to a database
The sheet becomes a nightly CSV. Nobody types into it as the source of truth after go-live.
- 06
Keep the original file next to the row
When it is wrong, you need evidence, not an argument.
- 07
Add a customer login only when they need status
That is a portal. Scope it. Do not bolt ten widgets onto the sheet instead.
- 08
Fire the new path on a Friday as a test
If Monday cannot run, you bought a boss. Undo it.
Keep these
- AI is optional on top of a table. The table is not optional.
- You own the repo on final payment.
Do this before you spend
- 01Export the sheet and delete zombie columns.
- 02Name who is allowed to confirm a row.
- 03Write the one sentence: after go-live, the sheet is an export.
This is Invent: $10–25k, one system of record, a human lock if a clerk drafts the row, the repo on final payment.
Questions
- When should you replace a spreadsheet with custom software?
- When two people cannot edit the same sheet without breaking a job, when a customer should see status, or when the sheet has become the company.
Next how-to
- How to choose custom vs off-the-shelfA decision test: buy when the job is common, build when the workaround is the company, sequence a migration so you are not betting the week.
- How to go from prototype to productionProduction means auth, real data, a kill switch, and a URL the company can send. The path in between, without a two-year programme.
- How to pick the first AI jobCircle work that is frequent, textual, and cheap to get wrong. Publish the truth first. Then a clerk — not a strategy.
Next
Replace the sheet with a system you own.
Invent is $10–25k for the queue, the confirm step, and the export. Fixed scope. Nothing billed before you sign.