All-Source Intelligence

All-source intelligence is analysis that fuses every available collection discipline into a single assessment: signals intercepts, human sources, satellite imagery, open-source material, measurement data, whatever exists. The term distinguishes the analyst who weighs all of it from the single-discipline specialist who produces one stream. A SIGINT report tells you what was said on a radio. The all-source analyst decides what it means next to the imagery, the agent report, and the unit's known doctrine.

The reason the discipline exists is that single sources mislead. Each collection method has characteristic blind spots and characteristic ways of being fooled. Radios can broadcast deception. Agents lie or get played. Imagery shows a convincing decoy tank at the resolution most satellites deliver. The 2003 Iraq WMD estimate is the canonical failure study: thin sourcing, most infamously the fabricator codenamed Curveball, propagated through assessments without the cross-source challenge that all-source practice is supposed to provide.

How the fusion actually happens

In US practice, all-source analysis lives at agencies like the CIA's Directorate of Analysis and the Defense Intelligence Agency, and in the military at every echelon's intelligence staff, the J2 or G2, where a fusion cell assembles the enemy picture that operations planning consumes. The workflow is unglamorous: collate incoming reporting against standing requirements, weigh source reliability, argue out contradictions, and publish an assessment with explicit confidence levels. The phrase "we assess with moderate confidence" is the discipline's signature, and it is doing real work, encoding how good the sourcing was.

The current pressure on the field is volume. Commercial satellite imagery, social media, and intercepted electronics produce more raw material in a day than Cold War agencies saw in a year, which is why every major intelligence organization is investing in machine-assisted triage, and why the analyst job is drifting from finding scarce data toward disbelieving abundant data.

How is this different from OSINT?

Open-source intelligence is one input, the publicly available stream. All-source is the layer above every input. The distinction got blurrier when open sources became rich enough to carry assessments on their own. Ukraine-war OSINT analysts geolocating strikes from TikTok videos do genuinely useful work, but they see only what is public. The all-source analyst's advantage, and burden, is holding the classified and unclassified pictures together and noticing where they disagree.