> For the complete documentation index, see [llms.txt](https://raic-labs.gitbook.io/raic-fed/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://raic-labs.gitbook.io/raic-fed/getting-started/01-platform-overview.md).

# What is RAIC Fed?

Mission imagery multiplies. Analyst time does not.

Organizations processing large-scale geospatial data are collecting more than ever: satellite passes, drone feeds, and multi-domain sensor data. The volume is staggering. And the teams tasked with making sense of it are still working with tools that were not designed to scale.

RAIC Fed is built to close that gap.

***

## The core capability

RAIC Fed enables broad-area imagery search, object organization, and analyst-driven model building, starting from a single seed image or hand-drawn sketch. What traditionally required a team of AI specialists, months of labeled data, and expensive infrastructure can now be done by an Analyst, in minutes, on the data they already have.

> **Note** RAIC Fed is not a replacement for analyst judgment. It is a force multiplier, keeping the Analyst in control while dramatically increasing the ground they can cover.

***

## What the Analyst can do

* **Search millions of square miles** across large imagery datasets using a single reference image as the starting point.
* **Organize results into a shared taxonomy:** a structured, evolving knowledge base that grows with every mission.
* **Train shallow models** on a small number of labeled examples, without waiting on an AI development cycle.
* **Surface unknown activity and monitor threats continuously**, reducing missed signals at scale.

***

## How it works: the analyst loop

RAIC Fed is built around a continuous, analyst-driven workflow:

1. **Search.** The Analyst runs a broad-area search using a seed image. Results are distributed across days, locations, and areas of interest, giving immediate situational context.
2. **Organize.** Results are structured into a shared taxonomy. The Analyst separates similar objects into distinct categories based on visual characteristics, context, and mission relevance.
3. **Refine.** Approved images are retained, rejected ones removed, and detections reassigned as understanding improves. The taxonomy grows into a persistent, trusted knowledge base.
4. **Model.** Once a taxonomy category is established, the Analyst applies a shallow model to that category. Unlike traditional pre-trained models, shallow models require only a few labeled examples and minutes to train.
5. **Iterate.** New results are validated and fed back into the taxonomy, continuously improving search precision and building mission-specific knowledge over time.

This loop turns raw imagery into reusable operational knowledge, compounding over time with every search the Analyst runs.

***

## Where RAIC Fed runs

RAIC Fed runs where sensitive data already lives. No re-architecture required.

| Deployment      | Description                                                                                                               |
| --------------- | ------------------------------------------------------------------------------------------------------------------------- |
| **Cloud**       | Hosted environment for unclassified or commercial workflows.                                                              |
| **On-premises** | Deployed within the organization's own infrastructure.                                                                    |
| **Air-gapped**  | Fully isolated environment for classified operations. IATT achieved; SECRET environment live; Top Secret ATO in progress. |

***

## How RAIC Fed is different

Traditional analytics pipelines are brittle, expensive, and slow to adapt. RAIC Fed shifts the model.

|                          | Traditional pipeline                              | RAIC Fed                                                    |
| ------------------------ | ------------------------------------------------- | ----------------------------------------------------------- |
| **Infrastructure**       | Centralized cloud                                 | Containerized; runs where data lives                        |
| **Security**             | Unclassified only                                 | IATT achieved; SECRET live; TS ATO in progress              |
| **Model training**       | Requires AI specialist and large labeled datasets | Analyst-driven; few labeled examples; minutes to train      |
| **Taxonomy**             | Fixed categories tied to a single dataset         | User-structured; evolves across the full imagery collection |
| **TOI / AOI support**    | Data must be ingested as images                   | Native search and taxonomy across multiple collects         |
| **Workflow integration** | Limited                                           | API-first; built for enterprise workflows                   |

***

## Who uses RAIC Fed

RAIC Fed has two roles:

* **Admin.** Responsible for deploying and configuring the RAIC Fed instance, managing the environment, and maintaining access.
* **Analyst.** The primary user of the platform. The Analyst runs searches, builds taxonomies, trains Detectors, and interprets results.

> **Note** All users of a RAIC Fed instance share a single workspace. There is no separation between accounts; every Analyst sees the same datasets, taxonomy, and Detectors.

***

*Learn more at* [*raiclabs.com*](https://raiclabs.com)