.svg){kind=icon}
Learn
Processing, exploitation, and dissemination (PED) is the work of turning collected ISR data into usable intelligence and getting it to the people who need it: processing raw feeds into a readable form, exploiting them to extract meaning, and disseminating the result to decision-makers.
Processing, exploitation, and dissemination (PED) is the work of turning collected ISR data into usable intelligence and getting it to the people who need it: processing raw feeds into a readable form, exploiting them to extract meaning, and disseminating the result to decision-makers.
Why it matters to your mission. Collected ISR is worth nothing until it is processed, exploited, and delivered, and that pipeline is where intelligence most often arrives too late to act on. Accelerating each step, from raw feed to readable product to the decision-maker, buys back time. The outcome is usable intelligence in the right hands while the decision window is still open.
Collection gets the data. PED is what makes it matter. A sensor that captures an image, intercept, or signal has produced nothing a commander can act on until that material is processed, exploited by an analyst, and delivered in time. PED is where most of the delay, and most of the backlog, actually lives, and it is the stage the Department of War is under the most pressure to accelerate as sensor counts climb faster than the analyst-hours to work them.
PED is the set of functions that convert collected data into intelligence products. It sits inside the joint intelligence process, which moves from planning and direction, through collection, processing and exploitation, and analysis and production, to dissemination and integration. In the ISR community, "PED" names the processing, exploitation, and dissemination of what sensors collect, often run by a dedicated PED cell staffed to work a specific feed or sensor.
The reason PED is named apart from collection is that the two are funded, manned, and constrained separately. A command can buy more collection and still be unable to use it, because the people and tools that process and exploit it have not scaled with the sensors. The Chairman of the Joint Chiefs of Staff governs this work in JP 2-01, Joint and National Intelligence Support to Military Operations (5 July 2017), which sets the doctrine for how joint forces process, exploit, and disseminate what they collect.
The three words are distinct jobs, and a gap in any one breaks the chain:
Processed data no one exploits is wasted collection. Exploited intelligence no one receives in time is a missed decision. The discipline of PED is keeping all three stages moving at the pace the mission sets, not just the stage a given team happens to be good at.
TCPED (tasking, collection, processing, exploitation, dissemination) is the ISR cycle that frames PED with the two steps that precede it: tasking the sensor and collecting the data. The broader joint intelligence process in JP 2-01 adds analysis and production and an evaluation loop around the whole sequence. PED is the shared middle in all three framings: the conversion of collection into product. The terms overlap because they describe the same work at different scopes.
| Framing | What it spans | Where PED sits |
|---|---|---|
| PED | Processing, exploitation, dissemination | The whole thing |
| TCPED | Tasking and collection, then PED | The last three of five steps |
| Joint intelligence process | Planning/direction, collection, processing/exploitation, analysis/production, dissemination/integration, with an evaluation loop | The processing/exploitation and dissemination phases |
Reading the table top to bottom, the wider the frame, the more context PED gains, but the core job never changes. If you remember one thing, it is that PED is the conversion step; the surrounding labels only decide how much of the lifecycle you name alongside it.
Because collection has outrun the capacity to exploit it. Doctrine is explicit that collection requirements routinely exceed available collection and PED capacity, so teams cannot process and exploit everything they gather, and prioritization is a standing part of the job rather than an exception. See JP 2-01. The constraint is rarely the sensor. It is the analyst-hours to process and exploit what the sensor already captured.
The same shortage reappears one layer down when a command tries to automate its way out of the backlog. The U.S. Government Accountability Office found that acquiring and curating usable, labeled data is a central challenge in fielding AI for defense; its report notes, for example, that detecting an adversary's submarines "requires gathering many images of various submarines and labeling them so the AI can learn to identify one on its own." See GAO-22-104765 (2022). The exploitation bottleneck does not vanish when you add automation; it moves into the data pipeline that feeds the automation. That is why a credible PED modernization plan treats data readiness and auditability as first-order problems, not afterthoughts.
Why PED is the bottleneck
Collection has outrun the capacity to exploit it. The constraint is rarely the sensor; it is the analyst-hours to process and exploit what the sensor already captured.
Doctrine (JP 2-01) is explicit that collection requirements routinely exceed available collection and PED capacity, so prioritization is a standing part of the job rather than an exception.
AI moves the routine conversion work off the analyst so judgment goes where it matters. It detects and labels objects, correlates reports across sources, and formats output for dissemination, with the analyst accepting, correcting, or rejecting each result. The analyst stays accountable for the call, and every step keeps a traceable record, so a product can be defended later against the question of how it was derived.
The table below separates what the machine does at each stage from what the human still owns. The division is deliberate: automation carries the volume, the human carries the authority.
| Stage | What it does | AI decision-support | Human authority |
|---|---|---|---|
| Processing | Raw feeds to readable form | Decode, format, georeference, transcribe, and translate at scale | Confirms fidelity |
| Exploitation | Extract meaning | Detect, label, and correlate objects and activities | Accepts, corrects, or rejects each call |
| Dissemination | Deliver to decision-makers | Format and route products by need and classification | Owns the release decision |
The failure mode to avoid is full automation that removes the checkpoint, and the audit trail with it. For defense intelligence the audit trail is not optional, so decision support with a human in the loop is the design, not a limitation of it.
Torch.AI supports PED as government-owned reasoning infrastructure that complements existing collection systems rather than replacing them. NEXUS makes multi-source ISR output meaningful through semantic representation and Knowledge Domains, so processed material from different sensors and formats can be reasoned over as one body of evidence. HALO makes the relationships among entities, events, and sources understandable through graph-based fusion and reasoning, which is the connective work that exploitation depends on. Both run on ORCUS, the government-owned data-availability layer, so the infrastructure stays owned by the mission and every derivation remains inspectable.
Because the infrastructure is government-owned, the exploitation logic and the provenance of each product stay with the customer rather than inside a vendor's proprietary environment, which is what makes a PED product defensible after the fact. See how this is packaged on the Torch.AI software page, and the companion piece AI-Enabled ISR and PED Automation for how the automation works across the ISR cycle in practice.
Evaluating a PED capability? Bring these seven requirements to a scoping call and we will walk each one against your environment and a sample of your collection: request a technical walkthrough.
PED is the set of functions that convert collected ISR data into intelligence products, named separately from collection because the people and tools that process and exploit data are funded and constrained separately from the sensors that collect it.
Processing converts raw collected data into a readable form; exploitation is the analytic step that extracts meaning from it; and dissemination delivers the result to decision-makers at the classification and in the form they can use, in time to act.
TCPED adds tasking and collection in front of PED, and the broader joint intelligence process adds analysis and production and an evaluation loop around it; PED is the shared middle, the conversion of collection into product, in all three framings.
Because collection routinely exceeds available collection and PED capacity, as joint doctrine states, so the constraint is the analyst-hours to process and exploit what sensors already captured, not the sensors themselves.
AI moves routine conversion work off the analyst by detecting and labeling objects, correlating reports, and formatting output for dissemination, while the analyst accepts, corrects, or rejects each result and every step keeps a traceable record.
Torch.AI supports PED as government-owned reasoning infrastructure, using NEXUS for semantic representation, HALO for graph-based fusion, and ORCUS as the data-availability layer, so the infrastructure complements existing collection systems and every derivation stays inspectable.
Describe the problem.
Define the constraints. We will structure the session.
No demo.
No slide deck.
Real work.
Describe the feed your team cannot keep up with. We will scope processing, exploitation, and dissemination support for it. Analyst in the loop. Your collection. Real work.
