The National Geospatial-Intelligence Agency unveiled a Rapid Capabilities Office at the GEOINT Symposium 2026 in Aurora, Colorado this week — a move that, if it follows through on its stated mandate, could make NGA one of the most accessible customers in the intelligence community for non-traditional vendors. The announcement came alongside a preview of an AI Blueprint that will define how NGA integrates artificial intelligence across its intelligence production mission, and a frank acknowledgment from leadership that the agency intends to "take a lot of risk" in acquisition.
What the RCO is designed to do
RCO Director Chris Parrett was direct in his description of the office's philosophy at the symposium: "We're not going to wait on the perfect solution." The RCO will evaluate technology through "shorter duration pilots and prototypes" rather than the multi-year formal acquisitions that have historically characterized NGA's procurement approach. This language maps directly to Other Transaction Authority (OTA) agreements, Cooperative Research and Development Agreements (CRADAs), and rapid prototyping vehicles — not traditional FAR-based contracts.
The comparison to Space Force's SpaceWERX, Army Futures Command's experimentation framework, and AFWERX is explicit: the RCO is NGA's attempt to build an innovation on-ramp that can evaluate and field commercial technology in months rather than years.
The AI Blueprint: what to expect
NGA Director Lt. Gen. Michele Bredenkamp confirmed at the symposium that an AI Blueprint is coming "soon." Based on her remarks and Via Satellite's reporting from May 6, the Blueprint is expected to address:
- Operationalizing AI across the intelligence cycle — moving from AI-assisted analysis to AI-native production of GEOINT products
- Replacing manual geospatial analysis workflows — specifically, the labor-intensive process of extracting intelligence from satellite imagery at scale
- Acquisition modernization — how NGA structures contracts for AI systems that need continuous model updates and retraining
- Data infrastructure — how NGA's classified data holdings are made available to AI systems under appropriate access controls
What this means for vendors
NGA has historically been one of the harder intelligence community customers to break into — dominated by SAIC, Leidos, Maxar, Planet, and a handful of other established geospatial contractors. The RCO is a deliberate attempt to change that dynamic. The signals for non-traditional vendors are unusually clear:
| Vendor Type | Opportunity | Entry Point |
|---|---|---|
| Computer vision / AI companies | Automated imagery analysis, change detection, object recognition | RCO pilot solicitations on SAM.gov |
| Satellite data analytics | Multi-source fusion, commercial data integration | NGA's commercial imagery procurement |
| Geospatial software firms | AI-native GIS tools, GEOINT workflow automation | GEOINT Pathfinder pilot program |
| Cloud / infrastructure | Classified compute for AI model training | IC GovCloud vehicles (AWS, Azure) |
The RCO headquarters: Reston, not Springfield
The RCO's location has not been confirmed in public reporting, but NGA's technology and innovation functions have increasingly been co-located with industry in the Reston/Tysons corridor rather than at the main NGA campus in Springfield, Virginia. This geographic positioning — near the venture-funded tech ecosystem of Northern Virginia — is intentional and consistent with how DARPA and DIU structure their industry engagement.
For small and mid-size firms, the practical advice is simple: watch NGA's SAM.gov postings and the GEOINT Foundation's industry day calendar for RCO-associated solicitations. The timeline from announcement to first solicitation in similar offices (SpaceWERX, AFWERX) has typically been 60–120 days.