Six ways to keep aircraft maintenance records
This compares approaches, not products — no vendor is named here, and none is being described sideways. We wrote the table about ourselves, so the useful part is where we put ourselves last. There are three of those, and they're called out below.
| Axis | Paper + a filing cabinet | A spreadsheet | Photos on your phone | A human digitization service | A flight-school ops platform | MyTailLog |
|---|---|---|---|---|---|---|
| Time to searchable records | Never — you flip pages | However long you spend typing | Instant to store, never searchable | Days to weeks, hands-off | However long you spend typing | Minutes per page, plus your review time |
| Cost | Free (a binder) | Free | Free | Per aircraft, in the hundreds — it is labour, so it can't be cheap | A monthly subscription | Free; there is no billing in the app at all |
| Who holds the data | You, physically | You — it's a file | You — it's a camera roll | You, once it's delivered | They do, on your behalf | You; MIT-licensed, so you can hold the server too |
| What you leave with | Nothing to leave | The file | The photos | Whatever was delivered to you | Whatever their export gives you | A ZIP of every row as JSON + the original scans, re-importable |
| If it shuts down | Doesn't apply | Doesn't apply | Doesn't apply | You keep the delivered files | You migrate on their timeline | Run the same code yourself |
| Answers questions, or just stores files? | Stores | Answers only what you built formulas for | Stores | Stores — accurately | Some: status and due lists | Answers: plain-English Q&A citing the entries, gap audit, equipment rebuilt from the logs |
| Tracks what's due | You and your A&P remember | If you built the formulas and keep them current | No | No — it's transcription, not tracking | Yes | Yes: Part 91 recurring items, ADs, email reminders with your own lead times |
| Per-entry correctness guarantee | The entry is the record | Exactly what you typed | The image is exact; nothing is structured | Strongest — a person read every entry | Exactly what somebody typed | Weaker — a model reads it, scores its own confidence, and you review it against the scan |
| Scheduling, dispatch, invoicing | — | — | — | — | Yes — that is the job | No, and not planned. Explicitly out of scope |
| Works with no signal | Perfectly | If the file is on the device | Yes | No | Usually web-only | iOS app (TestFlight beta) browses everything offline and captures offline; no Android app |
| Open source / self-hostable | Doesn't apply | Doesn't apply | Doesn't apply | No | No | Yes — MIT |
| Is it the legal record? | Yes — this is the record | No | No | No | No | No — an index of the paper, never a replacement for it |
Where we lose
A comparison table whose author wins every row is an advertisement. These are the rows we don't win, stated in full rather than buried:
- A human beats a model on per-entry certainty.Paying someone to read every entry gives you a person's guarantee on each one. We give you a confidence-scored draft: a percentage on every field, the scan cropped to where each value was read from, and the flagged ones held back from bulk-confirm. That catches what the model knows it's unsure about — it cannot catch what the model was confidently wrong about. Your review is what closes that gap, and it is real work.
- No scheduling, dispatch, or billing — ever.If you run a club, a school, or a leaseback and you need a booking calendar, tach-to-invoice dispatch, rental rates, or membership dues, an ops platform does that and we deliberately do not. It is on the out-of-scope list, not the roadmap, because building a second-rate scheduler would cost the records depth that's the entire point of this.
- Mobile is a beta, and iOS only. The native app is in TestFlight: it browses every entry, document and scan fully offline and captures pages offline, but editing existing entries is still done on the web, and there is no Android app.
When to pick something else
- An ops platform — if the thing keeping you up at night is who has the plane on Saturday and whether the renter got billed.
- A digitization service— if you would rather pay money than spend evenings reviewing, and you want a person's name attached to every entry.
- Paper and a binder— if you fly a hundred hours a year, one book covers it, and you can find what you need in it. It is still the legal record either way, and there's no shame in that being the whole system.
- The spreadsheet you already have — if it already answers your questions. Switching costs you the evenings; only do it if the answers are missing.
Where this approach is genuinely different
One thing, mostly: what happens to a record after it's captured. A photo is storage. A transcription is accurate storage. Here, extraction is the beginning — the entries feed a searchable cross-logbook timeline, a status grid, a Part 91 forecast, AD/SB compliance with FAA reference lookup, an equipment list rebuilt from what the logs say was installed and removed, weight & balance revisions, oil wear-metal and consumption trends, a records-gap audit, and plain-English questions answered with the source entries cited.
The second thing is the exit. MIT license, a full ZIP of your rows and original scans on demand, and a read-only public API you consent to per-app. Portability that survives us losing interest is a stronger promise than portability that depends on us staying nice.
The row that resets the table
Only one column answers “yes” to is it the legal record, and it's the filing cabinet. Under 14 CFR 91.417 the physical logbooks are the system of record, and every other column here — ours included — is an index, a convenience, or a search tool sitting on top of them. Whatever you choose, keep the paper.
More on how the pieces work in Help & documentation, and on cost, privacy and export in the FAQ.