Methodology & AI Limitations

Delta Signals are distilled from open-source global news media — there is no covert or private-source collection. The pipeline is automated and runs continuously:

1. Ingest — worldwide reported events are pulled from open news-event data on a rolling basis.
2. AI enrichment — each event is classified (category, escalation trend) and given a severity and a GEOINT-relevance score.
3. Satellite assessment — events that clear a relevance threshold are additionally assessed for whether a satellite could observe a physical mark, with a recommended sensor and collection window. This step is deliberately not run on every event: it is the most expensive part of the pipeline, and most of what global news reports (procedural, legal, business and retrospective items) has no physical footprint to image. Signals that were skipped are marked as such rather than being reported as un-observable — the API distinguishes the two with assessment_state (assessed / not_assessed) and assessment_skip_reason. Treat a missing satellite recommendation as “the question was not asked”, not “the answer was no”.
4. Geolocation — the event is resolved to coordinates and a country, corrected from the raw feed where possible so the mapped point matches the described location.
5. Source-linking — every signal keeps links to the original reporting so the evidence is one click away.

The Daily World Brief and the Delta Agent analyst are produced on top of this same corpus.

Each signal carries several distinct scores — they measure different things and should not be collapsed into one number:

• Event Severity (1–10) — the scale of the physical harm and footprint of a new event, not how newsworthy or how morally serious it is. This distinction is load-bearing: an individual crime, an arrest, a court ruling or a political announcement is scored 1–3 by design however grave it is, because it has little physical footprint. A high-severity event is also not necessarily observable from space.
• GEOINT Relevance (1–10) — how useful the event is for geospatial/satellite analysis. It is a different question from severity, but it is capped by it: severity 1–3 limits GEOINT relevance to 4, severity 4–5 limits it to 6, and only severity 6+ can reach 10. So a low-footprint event cannot come out highly GEOINT-relevant, by construction.
• Collection Priority (0–100) — the continuous ranking to use when deciding what to image. GEOINT Relevance is a coarse 1–10 integer under the cap above and saturates at the top, so it separates events poorly; Collection Priority combines imageability, tasking value and geo-readiness and is what the sort=geoint feed ranks by. Prefer it over the raw GEOINT score for prioritisation.
• Satellite Observability — three values, not two: observable, not-observable, and insufficient-detail (the reporting does not say enough to judge). The third is the common case and is not a soft “no”. A separate field states whether a satellite observation actually validated the mark (observability_evidence_status: physically-validated / claimed-observable / not-validated) — a described mark and a confirmed one are different claims, and only the first is common.
• Geolocation Confidence — whether the mapped point was geo-checked and is consistent with the resolved country (API field geo_verified). Do not read that field on its own. The check only ever asked “is the coordinate inside the stated country”, which is satisfied automatically when the coordinate is the country or region centroid — so geo_verified: true can sit next to a point that localises nothing. Two fields beside it say how to read it: geo_verified_scope (site / admin1 / country / none) and geo_verified_meaningful, which is false exactly when the check established nothing. geo_precision states the resolution granularity, and the published coordinate is rounded to match it.
• Source Confidence — how well-corroborated the event is (number of independent sources and mentions).
• Expected Information Gain (0–1) — how much a satellite look could reduce uncertainty, used to prioritize what is worth imaging.

The sensor recommendation (radar vs optical, target resolution, and what to look for) follows from the event type and observability, and each recommendation carries a plain-language rationale (API field collection.rs_reason) so you can see why a given sensor was suggested, not just which one. Ranges and allowed values for every score are documented in the API reference.

Everything below is produced by a large language model:

• Signal enrichment — category, severity, GEOINT relevance, escalation trend, geolocation correction, market-impact reasoning, and satellite-collection recommendations.
• The Daily World Brief — a synthesized narrative over the day's signals, produced under quality checks (see below).
• Delta Agent — an agentic analyst that reasons over the corpus and returns a sourced brief.
• Signal assessment — the per-signal remote-sensing deep dive (assess_signal), which reasons about what imagery could confirm or refute a specific event.
• Location refinement — refine_location re-reads the source reporting to tighten a signal's coordinate. It is worth stating plainly that a model, not a gazetteer lookup, is what moves the point; the returned precision and the reasoning are exposed with it.

Deterministic code — not a model — decides everything that gates or prices a result: plan entitlements, token charges, the count of independent sources, the quality gates below, and collection-readiness. Those checks also constrain model output rather than trusting it: contradictory enrichment fields are suppressed before they reach you, and the satellite-observability verdict a model writes is re-projected onto a fixed value set. The verdict itself still originates with a model — what is deterministic is the gating, not the judgement.

Enrichment is grounded in the source reporting rather than the model's own memory, which reduces (but does not eliminate) fabrication. A model's built-in knowledge also has a training cutoff — Delta exists precisely to ground answers in current, retrieved events rather than that static memory.

By default, no. Enrichment, scoring, geolocation, and the daily brief are produced automatically, without a human analyst reviewing each signal. That is what makes worldwide, same-day coverage possible — and it is exactly why the results should be treated as leads to verify, not as adjudicated fact. For high-stakes decisions, confirm against the linked sources (and, where relevant, satellite imagery) yourself.

• False positives / noise — an event may be mis-categorized, over- or under-rated on severity, or duplicated across reports.
• Mislocation — geolocation can resolve to the wrong place, especially for same-name locations; a signal may carry a lower geolocation confidence or land at country granularity only.
• Source-dependent — coverage and framing inherit the biases and gaps of the underlying news reporting; absence of a signal is not evidence that nothing happened.
• AI error — scores and summaries are model estimates and can be wrong; the source links are the ground truth, not the generated text.
• Observability is a recommendation — the satellite-collection suggestion is an assessment of what could be imaged, not a guarantee of a usable image (weather, revisit, and licensing still apply).
• Market tags are informational — any market-impact tag is context, not investment advice.

Scores and enrichment are validated against allowed ranges and value sets, and the Daily World Brief passes a deterministic quality gate before publishing: a truncated or structurally-incomplete brief is held back and the previous good brief keeps serving, rather than publishing a half-finished one. You can check current data freshness and pipeline status at any time via the API (GET /api/v1/status) or the MCP status://current resource.

Every signal links to its original reporting; when you export or republish, keep those attributions. Satellite imagery shown or searched in Off-Nadir Delta comes from open programs (e.g. Copernicus Sentinel-1/2), which are free to use with attribution to the providing agency. Follow each provider's attribution and licensing terms when you redistribute derived imagery or products.

Use Delta to find, prioritize, and orient — what is happening, where it is concentrating, and what to look at next — then verify the specifics against the linked sources before acting. Do not treat a single signal, score, or generated summary as confirmed fact, and do not use the service for unlawful surveillance or targeting of individuals.