Help & documentation

MyTailLog is a searchable index of your physical logbooks and a decision-support layer on top of them — not the legal maintenance record and not an airworthiness determination. The paper logbooks remain the system of record (14 CFR 91.417).

Almost everything you see is derived — read from your scans by AI, or computed from data you entered. Values can be wrong. Treat MyTailLog as a fast way to find, cross-check, and forecast — then confirm anything important against the paper before you rely on it. Low-confidence extractions are flagged for exactly this reason.

1. Explore the demo — every account gets a read-only demo aircraft (N734DM) on the dashboard, so you can poke at Status, Timeline, and Ask before scanning anything of your own.
2. Enroll an aircraft — the FAA registry lookup fills make/model/serial from the tail number. Five logbooks are created automatically: airframe, engine, prop, avionics, and Other.
3. Capture or uploadyour logbook pages (and A&P documents into Other).
4. Extract — AI reads each page into structured entries.
5. Review — confirm or correct what it read.
6. Then Status, Maintenance, Compliance, Ask your logbook, and the rest come alive from your data.

A native iPhone / iPad app is in TestFlight beta. It syncs your aircraft once, then works fully offline — which matters, because the hangar is exactly where you have no signal.

Statusanswers “can I fly this?” without a connection: current tach and hobbs with where each reading came from, then every maintenance item and recurring AD worst-first, with the same countdowns and calendar-month rules the web app uses. The aircraft list flags the worst one, so you see a problem before you open anything.

Recordwrites back. Type the meters you're looking at (pre-filled, and it warns if the number is lower than the last one), log oil you added, raise a squawk, or mark a recurring item done. A VOR check asks for the place, the bearing error and your signature, because 91.171(d) requires those recorded and signed — so it resets the counter and writes a real log entry, rather than just moving a due-date.

Everything you record is saved on the device immediately and uploads on the next sync. A Waiting to upload list shows anything still queued, and keeps the reason on anything the server refused — nothing you record disappears quietly.

Documentskeeps your AROW paperwork to hand — airworthiness certificate, registration, POH/AFM and weight & balance — for a ramp check with no signal. PDFs open in the app, page by page, which matters because a registration or airworthiness certificate usually is one.

Two ways to get pages in, both landing in the same review queue:

• Capture— opens your phone's own camera app, so you get its autofocus, HDR and stabilisation, and the shot is instant. Fill the frame with the page; it doesn't need to be cropped or perfectly square, because the extractor reads the page out of the photo. Blurry or glare-heavy shots are flagged before you keep them. Got the desk in the shot? Open the page's review screen and use Crop & clean up — drag the corners to cut out the background, rotate, and turn on the scan look for a black-and-white document finish. It replaces the stored scan(your paper page is still the legal record), so re-extract afterwards if the crop changed what's readable. On iPhone the app uses Apple's own document scanner and does this as you shoot.
• Upload scans — PDF, JPEG, or PNG. Multi-page PDFs are split into one page each, in order.

Already have the history as data rather than paper? Import a CSV instead — no scan needed.

Uploaded a logbook out of order — the second volume before the first? Open that logbook on Aircraft → Pages, sort by Entry date, and hit Save this order. It renumbers that logbook's pages to match what you're looking at, in one step. (It only offers this once every page in the logbook has been extracted — before that there are no dates to order by.) The arrows in Reorder still nudge a single page up or down.

Easiest way in: open Aircraft → Pages, tap the logbook you're about to scan, and use Capture into <that logbook>— the camera opens with it already selected, so there's no dropdown to get wrong.

Pick the right logbookfor what you're scanning — airframe/engine/ prop/avionics for running maintenance pages, and Otherfor A&P documents (see below). Pages queue on-device and upload when you're online, so capture works in a hangar with no signal.

Note: scanned logbook spreads sometimes contain two physical pages side by side — the extractor flags these so you can review both halves.

Coming from another platform, or from your own spreadsheet? Export it as CSV and bring the maintenance history in directly — no printing to PDF, no re-typing. Up to 5 MB / 5,000 rows per file; comma, semicolon and tab separated files all work, as do quoted fields containing commas or line breaks.

We don't ship an importer per vendor. Instead, the columns are mapped, not the rows: one AI pass reads your header plus a few sample rows and proposes what each column means (“Tach Out” → Tach, “A&P” → Signature, “Invoice #” → don't import). You confirm that mapping once, and every row is then converted in plain code — so the same file always imports the same way, and a wrong import is explainable by pointing at the mapping. Pick which logbookthe file belongs to; a spreadsheet doesn't say.

Dates are never guessed. 03/04/2026 is 3 April or 4 March depending on who exported it, and getting it wrong would shift a maintenance date by up to eleven months. We scan the whole date column: the first date anywhere in the file with a day past the 12th settles the reading for the entire file. Only if everyrow reads both ways are you asked — and you're shown what the first few dates become each way. A column that's internally inconsistent (some rows only day/month, others only month/day) is reported as a broken file rather than coerced.

Before anything is written you get a count of what will be created, plus every row that can't be read and why — an unreadable date or an implausible tach fails that one row rather than importing as a zero, and never takes the rest of the file with it.

Extraction reads a page image into structured entries (date, hours, work performed, parts, AD/SB references, signature). Each field gets its own confidence score, shown as a percentage right on that field in the Review screen — anything below threshold is highlighted. Beside each field you also get a cropped snippet of the scan showing where that value was read from, so you can confirm it without hunting the whole page (snippets appear once a page is extracted under the current model; re-extract older pages to get them). The full page image sits alongside the entries too. Editing an entry marks it confirmed.

Stickers at odd angles.Shops stick labels wherever there's room, so one page can carry an upright sticker and another rotated 90°. Each sticker is normally its own entry, and a page is read in every orientation. If rotated content is spotted but can't be read in full, the page gets a second look automatically — and if that still doesn't resolve it, Review shows an amber warning on that page asking you to compare against the paper. That matters more than a wrong value does: a low confidence score flags a field you can see, but an entry that was never extracted has nothing to flag. Use Add an entry the extractor missed to key it in. You can re-extracta page (e.g. if a multi-page entry wasn't linked) right from the review screen — it replaces that page's entries. The Logbook pages view (in the left nav) lists every captured scan grouped by logbook with its Needs review / Processing status, and can be sorted— within each logbook — by upload order, entry date, or tach, ascending or descending (the choice is remembered). Sorting reorders pages inside their own logbook, never across logbooks; date/tach come from a page's extracted entries, so an early logbook you uploaded late still sorts into chronological place. Filter to a single logbook and hit Reorder to hand-arrange its pages with the ↑/↓ arrows — handy for scans captured out of order. The aircraft Overview summarizes how many pages still need review.

To move faster, Review all (in the left nav) puts every extracted entry in one scrollable list — edit inline and confirm as you go, or hit Confirm N clean to accept, in one click, every entry the AI was fully confident on (high overall score, no flagged field, not a page-spanning fragment). Anything with a low-confidence field is left for you; use Open page ↗ on any group to check it against the original scan.

Re-uploading, re-capturing, or re-extracting the same page can leave duplicate scans and entries behind. Duplicates & fixes (in the left nav, editor-only) flags likely duplicates by matching on date, tach/hobbs, and work text — grouping pages that look like the same scan and entries that look like the same logged event. One copy in each group is marked suggested keep; delete any of the others. Deleting a page also removes its entries. Matches are heuristic, so review each group against the source scan (each row has an Open link) before deleting.

It also flags mis-keyed hobbs/tach readings — a value below an earlier reading (a dropped digit, e.g. 303 for 3,302) or an unusually large jump (a fat-finger, e.g. 33,303) — and suggests the corrected number. Accept it, edit the value first, or Keep mineif it's right; either way it won't re-flag. This covers both log entries and MyFlightBook-synced readings. It also catches a Hobbs value that's just the Tach duplicated into both fields (they should differ) — offering to Clear the bogus Hobbs while keeping the Tach.

Scan into the Otherlogbook for documents an A&P shop produces that aren't running-log pages. MyTailLog classifies each and applies it instead of storing it as log entries:

• Weight & Balance sheet → creates a new W&B revision(empty weight, CG, moment), which becomes your current W&B.
• AD compliance report → treated as the ground truth for your AD state. It corroborates matching tracked ADs (they get a “✓ A&P report”badge), and creates tracked records for ADs it lists that you weren't tracking yet.

Every extracted entry across all logbooks, merged into one date-ordered timeline and filterable by logbook type. Full-text search finds work by keyword, part, or AD number (e.g. “oil change”, “magneto”, “AD 2015-19-07”). Each entry links back to its source page image.

Ask plain-English questions (“When was the last annual?”, “Hours since prop overhaul?”). Answers are drawn only from your extracted entries and cite the specific entriesthey came from, so you can click through and verify. It won't invent facts the entries don't contain — but, like all AI, it can misread, so confirm anything you act on.

The at-a-glance dashboard: every recurring inspection, maintenance item, and AD as a color-coded card — red = overdue, amber = due soon, green = current— with days and hours remaining — plus, where there's enough flying history, an approximate calendar datefor the hours to run out. It's a read-only rollup of your maintenance forecast and recurring ADs; each card links to where you manage it.

Tracks recurring Part 91 items (annual 91.409, transponder 91.413, pitot-static 91.411, ELT 91.207, VOR, 100-hour, oil, TBO, prop overhaul) and computes next-due from the interval and last-done. Seed the standard set with one click, mark items done, or add your own. Regulatory items are distinguished from advisory ones (TBO/overhaul).

Calendar months, not exact months. The regulations that count in months — annual (91.409), transponder (91.413), pitot-static (91.411), ELT (91.207) — say “within the preceding N calendar months”, so the day of the month the work was signed off doesn't matter: an annual completed anywhere in March 2025 is good through 31 March 2026. Due dates for those items land on the last day of the month. Advisory items (prop overhaul) and the VOR check are not calendar-month rules — 91.171 is 30 days — so those count from the date itself.

“Due in 38.4 hours” is accurate but hard to plan around. Wherever an item counts down on hours, MyTailLog also shows an approximate calendar date — ≈ 14 Mar 2027 — worked out from how much this aircraft has actually been flown, using your own logged meter readings from the past 365 days.

It is a planning estimate, not a determination. The hours figure is the thing that comes due; the date is only our arithmetic on a rate that can change the moment your flying does. Every projection carries the ≈ sign and a confidence level, and hovering it shows the rate and the exact readings it came from, so you can judge it yourself.

Confidence reflects how much evidence sits behind the rate — both how many readings and how long a stretch they cover:

• High — 6+ readings spanning 180 days or more.
• Medium — 4+ readings over 90 days.
• Low — 2+ readings over 30 days. Treat as a rough sketch.
• None — anything thinner. No date is shown at all; you simply see the hours, exactly as before. MyTailLog will not guess a date from a single data point.

The rate is measured on the tach wherever possible, because hour-based limits — the 100-hour, TBO, hour-interval ADs — are written against engine time. If your logs carry no usable tach history it falls back to hobbs and says so in the hover text. Hobbs runs a little faster than tach, so a hobbs-based projection tends to land slightly early — the safe direction to be wrong in.

Aircraft → Meters & resetsshows what each instrument reads and lets you record two things the logbook alone can't express: a meter that was replaced, and airframe time.

A hobbs or tach that gets swapped restarts near zero. Record the replacement — the date, the old meter's final reading, and what the new one started at — and MyTailLog stitches the two together into one continuous total. Without it, importing older logbook pages makes time appear to run backwards: the boundary reading gets flagged as a typo, and any item last done before the swap counts down against a number it can never reach. Both meters are handled the same way, and an aircraft can have more than one replacement on file.

Airframe is a third, independent meter for sailplanes and motorgliders. A glider has no engine to drive a tach, and a motorglider accrues far more airframe time than engine time, so airframe is never derived from the other two — it shows only what has actually been recorded. Enter it on a logbook entry, at enrollment, or by hand on this page.

Where a reading came from. Each current reading shows the date it is anchored to and its source — a logbook entry, a MyFlightBook sync, something you entered by hand, an ADS-B estimate, or the starting reading. A value marked estimated was derived rather than read off the meter, and the source then names what it was derived from. The readings list labels each row the same way, so an ADS-B estimate is never mistaken for a meter read.

Starting readings on the same page are the figures captured when you added the aircraft. Every hours countdown starts from them, so a number typed into the wrong box shifts all of them — and they used to be set once at enrollment and unreachable afterwards. You can now edit them, and clearing a meter stops tracking it entirely: if you work from tach only, leave Hobbs blank and nothing will count against it. (The oil change is the one item that defaults to Hobbs; its “Hours counted on” setting switches it to tach.)

If you don't log every flight, the recorded hours drift below the real ones and every countdown reads optimistic. Aircraft → Meters can turn on a daily check against the OpenSky Network, a free community ADS-B feed, to notice that the aircraft flew when your records don't show it.

It is off by default and opt-in per aircraft. The only thing that ever leaves the app is the aircraft's ICAO 24-bit Mode S address — public FAA registry data. Nothing about you or your records is sent, and no track or position data is stored: only the start, end and duration of each flight seen. Turn it off and the checks stop.

The address is looked up from the FAA registry (or adsbdb as a fallback) and cached. It is never computedfrom your N-number — that encoding has edge cases that would silently pull another aircraft's flights. You can always enter it by hand.

The check runs once a day, at 17:30 UTC (1:30 pm US Eastern in summer, 12:30 pm in winter). So a flight this afternoon usually shows up tomorrow, not tonight. Each run looks back over the last three days, which means a run that fails or is skipped catches up by itself rather than losing those hours.

Your own records always win. Hobbs and tach are cumulative, so a reading from a MyFlightBook sync, a logbook entry, or one you typed already accounts for everything flown up to its date. ADS-B only speaks up about flights after your most recent reading — and it stays quiet entirely if a meter was replaced since then.

When it does speak up you get one line — “3 flights totalling ≈4.2 h since your last reading” — a suggested meter value, and two buttons: record a reading, or dismiss. Nothing is ever written for you. The suggested number is pre-filled and fully editable; read the real meter and correct it before saving.

Honest limits. Airborne wall-clock is neither tach nor hobbs — it excludes taxi and runup and drifts from tach with RPM. Ground coverage has gaps. Not every GA aircraft broadcasts ADS-B Out. Treat it as an estimate that prompts a real reading.

Track discrepancies noticed in flight — a rough mag, a sticky switch, a small leak. In Aircraft → Squawks, anyone with access to the aircraft (including a shared pilot) can report one with a severity; it stays open until an editor marks it resolved.

Owners periodically send an oil sample to a lab (Blackstone, AVLab, …) and get back a report of wear-metal concentrations. In Aircraft → Oil analysis, click Import oil reportand upload the lab's PDF (or a photo of it). The AI reads every sample in the report — dates, hours on the oil and engine, each element in parts per million, oil properties, and the lab's written comments.

Each wear metal is charted over time against the lab's universal average for your engine type (the dashed line). Values above it are highlighted amber, and well above (2×) red — a prompt to watch, not a verdict.

Separately, log each oil top-off(“added 1.5 qt”) with the tach/hobbs at the time — the meters prefill from the aircraft's current reading. From those, MyTailLog charts your burn rate (hours per quart) between top-offs — a leading indicator of engine health, distinct from the lab wear-metal trend above.

Like all extraction here, the numbers are read by AI — confirm anything important against the original report, and treat the lab's own assessment as authoritative. Re-importing the same report updates its samples in place rather than duplicating them.

Connect your own MyFlightBookpilot logbook to pull each aircraft's latest recorded hobbs and tach into MyTailLog. There is no app-wide account: in Profile → MyFlightBook you register your own OAuth app on MyFlightBook and paste its client ID and secret (the secret is stored encrypted server-side and never shown again), then click Connect and approve access. Once connected, Sync matches your MyFlightBook aircraft to MyTailLog aircraft by tail number and records the ending hours from the most recent flight.

Because matching is by tail, a shared aircraft can receive hours from any connected co-owner — whoever flew (and logged) most recently supplies the current reading.

Honesty caveat: MyFlightBook has no authoritative “current” meter for a shared plane, and a flight's ending hours may go unlogged. So this is the latest recorded hobbs/tach, shown as “as of <date>” — treat it as the last known reading, not a live gauge, and confirm against the aircraft.

Extraction and Q&A run on the app's selected AI provider. By default they use the app's shared key, which has a daily limit (both per person and an overall daily budget). If the shared budget for the day is used up, AI pauses until tomorrow. To bill AI usage to your own account and get a much higher limit, add your own Anthropic or OpenAI API key in Profile → AI & your API key. Your key is stored encrypted and never shown again; only the last four characters are kept for display.

The app uses the stronger model for handwritten logbook pages, where a missed digit or maintenance entry matters. Printed pages, PDFs, and Ask My Logbook use the provider's lowest-cost capable model. Marking a captured page as printed or handwritten controls that routing; an unknown page takes the safer handwritten path.

This must be an API key (starts with sk-ant-) from the Anthropic Console, where you add a little pay-as-you-go credit. A Claude.ai (Pro/Max) subscription won't work— it's a separate product with no API access.

With your own key connected, the panel shows your usage — calls, input and output tokens, and estimated cost so far(from token counts at the provider's list prices; a close guide, not your exact invoice). If your key is rejected (invalid or out of quota) the request fails with a clear message — it is never silently charged to the shared key.

A daily background check emails you beforemaintenance, inspections, and ADs come due — and once they're overdue. Turn reminders on and tune the lead times in Profile → Notifications. The master switch is the on/off for allreminder email; below it, each category sets how far in advance you're warned:

• Annual inspection — a number of daysbefore it's due (default 90).
• Oil change — a number of hoursbefore it's due (default 10).
• ADs / SBs and everything else — both a days-before and an hours-before window (defaults 30 days / 25 hours); whichever is reached first triggers the reminder.

Reminders are grouped into one digest email per day, organized by aircraft, each item linking to its Status page. You're reminded once per due-cycle: after you mark an item done and it schedules a new next-due, the reminder arms again for the next cycle. You never get an empty email.

Note: the oil-change interval(hours between changes) is set on the oil item on each aircraft's Maintenancepage — the setting here only controls how early you're alerted, not when the item is due.

Track Airworthiness Directives and Service Bulletins: compliance status (complied / previously complied / does-not-apply), method, dates, and recurring intervals → next-due. Look up the official FAA reference (Federal Register, with a DRS fallback) to confirm applicability. ADs referenced in your logs but not yet tracked are surfaced so you can add them.

Explore searches for ADs you might not know about, three ways at once: by manufacturer(the airframe make plus every installed equipment's make — the broad net), by model (your specific variant — the sharp one), and by any keywordyou type. Results come from the Federal Register plus the FAA's DRS, which is what surfaces the pre-1994 legacy ADsthe Federal Register archive doesn't hold; if either source is unreachable the other still returns. Each result lists the models the AD names, with yours highlighted and those results sorted first, so you can see whether your variant is actually named instead of guessing from the make. Track this AD adds it in one click as a one-time item, or Track with an interval records it as recurring on hours, calendar months, or both, with a next-due.

Search results are a starting point, not a determination. The parsed model list is read off the AD's title and summary and can be incomplete — real applicability often turns on serial numbers and installed equipment. Confirm against the AD itself; the call is yours and your A&P's. For ground truth, scan an A&P AD compliance report.

A reconstructed list of what's installed now, derived from your logs — installs and removals detected during extraction arrive as proposalsyou confirm or reject (nothing is trusted automatically). Use “Update from logs” to rescan the full history.

A history of your W&B revisions (empty weight, CG arm, moment) with the latest as your current W&B — plus useful load if you enter max gross. Enter any two of weight/arm/moment and the third is filled in. This is an index of your W&B records, not a loading calculator.

A home for the aircraft's permanent records— airworthiness certificate, registration, radio station authorization, POH/AFM, weight & balance, STCs, 337s, 8130-3s, ICAs, and manuals. In Aircraft → Records Vault, upload a PDF or photo (up to 25 MB), tag it with a category and reference number, and it's a click away instead of buried in a binder. These sit alongside your logbook scans.

Any Vault document can be attached to a specific maintenance entry — the 8130-3 for the part that entry installed, the 337 for the alteration it records. On an entry in Review, use + Link from Vault to pick an existing document (or + Add file to upload and attach in one step); unlink detaches it without deleting — the document stays in the Vault. The Vault shows the same link from the other side, as the linked record under each document, and attachments appear on their entry in the Logbook timeline. A document attaches to one entry at a time.

Editors can add, attach, detach, and remove documents; viewers can view and download them but cannot change any of it. Attachments travel in the .zip backup and survive a re-import.

Advisory heuristics that flag suspectedgaps in the digitized record: missing annuals (breaks in the annual chain), long stretches with no entries, and recurring ADs never complied or past due. Framed as “suspected” — a gap may just mean a page isn't scanned yet.

Two PDFs, both produced by your browser's Print → Save as PDF (nothing to install). The maintenance summaryis the one-page document you hand a buyer, an insurer, or your IA at annual: status at a glance, open squawks, the AD/SB compliance table, what's coming due, installed equipment, and current weight & balance. The full records report is the same thing plus every transcribed logbook entry. You can also download CSVs of entries / AD-SB / equipment / maintenance.

The .zip backup takes everything — all records plus your original scans — and includes a README.txt documenting every file and every column, so the archive still explains itself years from now.

Automatic cloud backups. Connect Dropbox and/or Google Drive from Profile and that same .zip is pushed to your own account monthly or quarterly, one file per aircraft, named <TAIL>/<date>-<TAIL>.zip. On Drive that sits inside a MyTailLogfolder; on Dropbox the app folder is already that folder, so there's no second one. You can connect both — each has its own schedule and runs independently, which is real redundancy rather than one basket. We can only ever see the files we put there, never the rest of your account: Dropbox gives us an app folder, and on Google Drive we ask only for permission to touch files this app creates, so your existing Drive is invisible to us by construction rather than by promise. We only ever addfiles — nothing in your account is renamed, replaced or deleted, so retention stays your call. Profile shows each destination's last run, its result and its size, and emails you if two runs in a row fail. Very large aircraft (hundreds of scanned pages) are reported as too large to upload rather than silently failing — download those by hand.

On lock-in. That .zip re-imports: restoring it recreates the aircraft, entries, ADs and scans as a new aircraft (it never overwrites an existing one). Everything inside is plain JSON and your original JPEG/PNG/PDF files — readable with no special software. MyTailLog itself is open source under the MIT licence and can be self-hosted against your own database, so leaving is always an option you actually have.

Invite others by email as viewers (read-only) or editors (can contribute) — they get access the moment they sign in with that address, no account needed first. Owners can transfer an aircraft to another user, or delete it (type-to-confirm; removes all records and scans). Access is enforced by the database, not just the UI.

Sign in with a magic link (no password) or set a password to use either. Manage your name, A&P/IA certificate number, email, and notification settings from Profile.

Other apps (like MyFlightBook) can read your aircraft's airworthiness, equipment, hours, oil, and weight-and-balance data — only with your consent, and read-only. On the consent screen you choose the scope: all your aircraft (the default — including any you add later, so an app keeps working as your fleet grows) or only the ones you pick. A brand-new account can authorize an app before adding any aircraft — the app just sees an empty list until you add one. Revoke any app anytime under Profile → Connected apps. Building an integration? Register an app and read the guide under Developer API. Your transcribed log entries are never shared.