The FAA rows are free and always will be. What's for sale is the validated pipeline and the discipline around it — and both offerings below are fixed-price, sized to move inside a fiscal year, and graded in writing against criteria set before any results exist.
Not data. The deliverable is a transferable capability: the audited ingest and enrichment pipeline, the 874-entry curated place gazetteer, the placebo and holdout validation machinery, and an integrity gate that is itself tested by corrupting the data 30 ways and requiring it to catch each one. The atlas at skywatch.analyticadss.com is the standing proof the discipline works — every figure on it regenerates from public data, and when we get something wrong it goes on a public retraction list, six entries long, two of which made the product look worse.
The build-vs-buy question deserves a straight answer. A capable data-science cell can reproduce this: recover the 2016–2019 quarters from archives, build the gazetteer, get the nulls right, and find the mistakes. We estimate that at multiple person-months, and the six retractions are the map of where that time goes — each one is an error a fresh replication would have to find the hard way. Buying the pilot is buying the already-audited version, with the error-discovery receipts published.
Two weeks, one named installation or site, buyable on a government purchase card with no procurement action. Deliverables:
If the site has fewer than 12 reports in a decade, we say so and deliver the honest version: what a thin record does and does not mean.
The same discipline pointed at your own incident data, inside your boundary. Your data never leaves your environment and we keep no copy — that is a contract term, not a slide. Structure:
Acceptance criteria, agreed at Day 0: the deliverables above on schedule; the fused view running on agency infrastructure; the grading published to you regardless of outcome. A draft sole-source justification is available on request, and source-and-data escrow with quarterly deposits is a standard term at no added cost.
| It is | It is not |
|---|---|
| A public-data incident atlas and a validated reporting baseline, placebo-tested against 499 population-matched controls | A sensor, a detection system, or any kind of live feed — the FAA record updates quarterly |
| A triage aid with a measured, modest edge: top flags about 2× more likely than chance to stay anomalous, walk-forward tested | A prediction engine — only about one top flag in seventeen is still anomalous a quarter later, which is why we say leads, not predictions, and print the number |
| A per-facility dossier and brief generator with the source reports attached | Attribution or intent — nothing in this data says who was flying or why |
| Reproducible and provenance-stamped: every figure traces to a data snapshot digest | Incursion counts — these are reports, and reporting moves with attention as well as activity |
Source code is proprietary commercial software under FAR 12.212. Work products produced for you under contract are offered with government-purpose rights by default. We claim nothing over government data — not the FAA rows, and not yours. The full deliverable-class mapping, including DFARS assertion tables on request, is in DATA_RIGHTS.md in the repository.
Analytica DSS is a small vendor, and the mitigation is engineering rather than reassurance: pinned dependencies, a one-command rebuild from public data, committed extraction caches, a runbook, and escrow as a standard term — built so that a competent engineer who has never met us could take the pipeline over. The continuity arrangement is named in the pilot terms.