aviationstack VS FlightFinder

FlightFinder's answer:

Most aviation tools tell you where a plane is right now. FlightFinder tells you what the official record says about it.

It joins corpora that normally live apart and in incompatible formats — NTSB and international investigation reports, the US Civil Aeronautics Board archive back to 1938, the FAA wildlife-strike database, NASA's ASRS confidential reports, airworthiness directives, service difficulty reports, runway incursions — into one searchable corpus, deduplicated across sources, with every figure traceable to the document it came from.

Two things follow from that:

• You can ask about your own flight. Enter a flight number and you get one page: that airline's record, the aircraft type's history, and a turbulence forecast for the route.
• The data is citable. Datasets are published openly with DOIs on Zenodo and a preprint on arXiv, and the API returns the same records the site renders.

There are no safety "scores" or star ratings for airlines. Aviation accidents are rare enough that ranking carriers on them is statistically dishonest, so the site shows counts, dates and sources and lets you read them.

FlightFinder's answer:

Deliberately boring, so the interesting part can be the data.

• Frontend: React with Vite, server-side rendered for the pages crawlers and search engines need to read without JavaScript.
• Backend: Node.js and Express.
• Database: SQLite via better-sqlite3, in WAL mode. Unfashionable for a few hundred million rows, but the workload is read-heavy and mostly analytical, and a single file that fits in page cache beats a network round trip.
• Mobile: Capacitor, so iOS and Android ship the same codebase as the web app.
• Infrastructure: a single Hetzner box behind Cloudflare, with pm2 for process management and Redis for rate limiting and shared caching.
• Data pipeline: the part that actually matters — scrapers, OCR for scanned government PDFs, and deduplication across sources. The scrapers are open source.

Turbulence forecasts are computed from NOAA GFS model output using the Ellrod index, not bought from an API.

FlightFinder's answer:

They solve a different problem, and for most of them FlightFinder is a complement rather than a replacement.

• FlightAware, Flightradar24 answer "where is this plane now" from ADS-B. Excellent at it. They do not carry accident, incident or airworthiness history.
• AirLabs, aviationstack sell schedule and status APIs to developers. Also excellent at it, and also not a safety corpus.
• The official sources themselves — NTSB, FAA, NASA ASRS — publish everything FlightFinder uses. But each lives in its own portal, its own schema, often as scanned PDFs, and none of them join to the others. Answering "has this aircraft type had problems" means visiting four sites and reconciling them by hand.

FlightFinder is worth using when your question is historical rather than live: what happened to this type, this carrier, this airport, and where is that written down. It is free to read, the sources are cited on every figure, and the underlying datasets are published with DOIs so you can check the work rather than take it on trust.

If you need live tracking, use a tracker. The two answer different questions.

FlightFinder's answer:

Four groups, in rough order of traffic.

• Nervous flyers with a specific flight booked. They are not browsing; they have a flight number and a date and want the honest answer before they board. The turbulence forecast and the per-flight brief exist for them, and the copy is deliberately calm rather than dramatic.
• Aviation enthusiasts and spotters looking up a registration, a type or an airline's history, and comparing aircraft families.
• Researchers, journalists and analysts who need the underlying records with citations — the open datasets, the DOIs and the API are aimed here.
• Aviation professionals — pilots, engineers, safety staff — checking airworthiness directives, service difficulty reports or ASRS narratives for a type they work on.

The common thread is that all four want the primary record rather than someone's summary of it.

FlightFinder's answer:

It started in March 2026 as a way to search flights by aircraft type — the enthusiast's question of "which flights are still operated by an A340". Answering it well meant assembling a proper aircraft and fleet database, and once that existed the obvious next question was what had ever happened to those aircraft.

That turned out to be the harder and more useful problem. Aviation is one of the most thoroughly documented industries on earth: every accident is investigated, every finding published, every airworthiness concern issued as a directive. But that record is scattered across dozens of government portals in incompatible formats, much of it as scanned PDFs, and none of it joined together. The historical material is worse — the US Civil Aeronautics Board reports from 1938 onward exist mostly as page images in archives.

So the work became extraction and reconciliation: OCR the scans, parse the tables, deduplicate the same crash reported by four different bodies, and join everything to the aircraft, airline and airport it belongs to. Then publish the cleaned datasets openly, with DOIs, so the extraction can be checked rather than trusted.

The flight-safety brief came last, and came from a simple observation: most people meeting this data are not researchers. They have a flight tomorrow and they are anxious. They deserve the same facts, presented calmly.