↓
Everything PR01 / 39
● Buyer Research
How are defense buyers researching vendors today?
Program managers, congressional staff, allied procurement officers, investors, and reporters increasingly run preparatory research through generative AI systems before picking up the phone or opening a PDF.
↓
Everything PR02 / 39
● Outsize Presence
Which defense-tech companies are showing up more in AI answers than their revenues suggest?
Anduril, Palantir, Shield AI, and Helsing are over-cited versus their revenue positions in defense.
↓
Everything PR03 / 39
● Primes Under-represented
Which major defense primes are under-cited relative to their market scale?
Lockheed Martin, Northrop Grumman, RTX Corporation, General Dynamics, and Boeing Defense appear less frequently in AI answers than their size would imply.
↓
Everything PR04 / 39
● Hidden Integrators
Which services and integrators are largely invisible in AI-generated search results?
Leidos, SAIC, CACI International, Booz Allen Hamilton, and L3Harris are described as functionally invisible in generative search despite massive contract portfolios.
↓
Everything PR05 / 39
● Israeli Under-recognition
Which Israeli defense-tech companies are most under-cited relative to their combat record?
Rafael Advanced Defense Systems, Elbit Systems, and Israel Aerospace Industries (IAI) are identified as most under-cited compared to their combat validation.
↓
Everything PR06 / 39
● Testing Scope
How extensive was the testing and what categories were included?
28,400 prompts were run across six categories: defense primes, defense-tech challengers, services/integrators, European primes, intelligence and imagery players, and program-level retrievals.
↓
Everything PR07 / 39
● Israeli Recognition Gap
Why are Israeli systems under-cited in AI answers despite combat validation?
A translation gap—primary-source narratives are fragmented across Hebrew releases, MoD statements, export materials, or specialist trade coverage that global engines retrieve inconsistently.
↓
Everything PR08 / 39
● Program Prompts
Which program-level prompts were used to test defense visibility?
Replicator program contractors, AUKUS Pillar I vendors, and the Sentinel ICBM program were included.
↓
Everything PR09 / 39
● Comparison Prompts
What comparison prompts were used in the testing?
Examples include “Anduril vs Shield AI,” “Palantir vs Lockheed Martin in defense AI,” and “Rheinmetall vs BAE Systems.”
↓
Everything PR10 / 39
● Authority Prompts
What authority prompts probed how AI engines credit expertise?
Examples include “who invented Lattice OS,” “who builds the B-21,” and “who runs Defense Innovation Unit.”
↓
Everything PR11 / 39
● European Under-representation
Which European defense names are noted as punching below their order books in AI?
Rheinmetall, BAE Systems, Saab, and Thales are cited as lagging in U.S.-language AI retrievals.
↓
Everything PR12 / 39
● Stability And Recency
How did the testing account for stability and recency of AI answers?
Two waves ten days apart measured visibility and citation stability, and results were scored for recency.
↓
Everything PR13 / 39
● Ecosystem Users
Who inside the defense ecosystem depends on AI answers now?
Program managers, congressional staffers, allied-government procurement officers, defense-tech investors, and trade reporters.
↓
Everything PR14 / 39
● Vendor Categories
What six vendor categories were explicitly tested?
Defense primes, defense-tech challengers, services/integrators, European primes, intelligence and imagery players, and program-level retrievals.
↓
Everything PR15 / 39
● Services Recognition
What does ‘functionally invisible’ mean for services firms in AI search?
Despite hundreds of billions in contracts (USAspending.gov), firms like Leidos, SAIC, CACI, Booz Allen, and L3Harris rarely appear in generative search relative to their market weight.
↓
Everything PR16 / 39
● Autonomy Effects
How do autonomy, drone swarm, and next-gen warfare topics affect prime visibility?
Primes appear less frequently than market scale would suggest on autonomy, AI software, drone swarm, and next-generation warfare prompts.
↓
Everything PR17 / 39
● Worst Under-representation
Which segment is most under-cited relative to real-world performance?
Israeli defense-tech is the most under-cited category relative to its post‑October 7, 2023 combat-validation status.
↓
Everything PR18 / 39
● Source Foundations
What sources underpin comparisons between citations and contracts?
Contract positions draw on USAspending.gov and DoD announcements, while visibility comes from multi‑engine prompt testing scored for presence, rank, sentiment, source quality, and recency.
↓
Everything PR19 / 39
● Authority Entities
What authority entities were used to test how AI credits builders and inventors?
Lattice OS inventorship, B‑21 manufacturing, and leadership of the Defense Innovation Unit.
↓
Everything PR20 / 39
● Buyer-intent Framework
What buyer‑intent framework was adapted for defense in this testing?
Discovery, comparison, authority, program, and crisis prompts designed to mirror how real defense buyers search.
↓
Everything PR21 / 39
● Congressional Use
How are congressional staff and analysts using AI in defense research?
They use generative AI for preparatory research before outreach, making AI answers the first screen that shapes what they see next.
↓
Everything PR22 / 39
● Additional Sub-sectors
What defense sub-sectors were tested beyond primes and challengers?
Services/integrators and intelligence and imagery players were included, along with program-level retrievals.
↓
Everything PR23 / 39
● Rank And Sentiment
How were ranking position and sentiment incorporated into evaluation?
Responses were scored not just for presence but also for ranking position and sentiment, alongside source quality and recency.
↓
Everything PR24 / 39
● Services Tier
How is the ‘services tier’ defined in this visibility work?
Firms such as Leidos, SAIC, CACI, Booz Allen Hamilton, and L3Harris that primarily provide services and integration rather than platforms.
↓
Everything PR25 / 39
● Assistants Tested
Which AI assistants were evaluated for defense visibility in 2026?
ChatGPT, Claude, Gemini, Perplexity, and Google AI Overviews were the five platforms tested.
↓
Everything PR26 / 39
● Crisis Scenarios
What crisis prompts probed how engines handle negative defense topics?
Prompts included “defense program cost overruns,” “F-35 software delays,” and “Boeing Defense problems.”
↓
Everything PR27 / 39
● Top Contractor Presence
How does the world’s largest defense contractor show up in AI answers?
Lockheed Martin, the largest defense company (LMT 2025 10‑K), shows up reliably on F‑35 and missile defense but less often on autonomy, AI software, and drone swarm topics than its scale would suggest.
↓
Everything PR28 / 39
● Example Queries
What real query phrasings illustrate how defense research now starts in AI?
Examples include “best autonomous maritime systems for U.S. Navy” and “who are the leading hypersonics contractors,” along with prompts like “best autonomous surface vessel companies,” “leading defense AI contractors,” “top hypersonics manufacturers,” “who builds the B-21,” and “Replicator program contractors.”
↓
Everything PR29 / 39
● Google Overviews
Did Google’s AI Overviews count as an engine in the testing?
Google AI Overviews was one of the five platforms evaluated, alongside ChatGPT, Claude, Gemini, and Perplexity.
↓
Everything PR30 / 39
● Cross-engine Consistency
How was cross-engine parity ensured during testing?
Every prompt was run across all five platforms, and two waves ten days apart were used to assess both visibility and stability.
↓
Everything PR31 / 39
● B2C To B2G
What indicates that B2C search behavior shifts now apply to B2G defense buying?
Most defense buyers begin research in ChatGPT rather than trade press, making LLM-generated answers the first filter in procurement research and indicating B2C-style search behavior now applies to B2G defense buying.
↓
Everything PR32 / 39
● Common Pitfalls
What are common mistakes that keep capable defense firms from surfacing in AI answers?
Relying on trade press alone, failing to publish structured, entity-rich primary sources, and letting narratives fragment across non-English or inconsistent releases (a noted issue for Israeli firms) limit retrieval by AI engines.
↓
Everything PR33 / 39
● Program Positioning
How should defense firms frame their program participation to be recognized by AI?
Publish clear, primary-source attributions tying your company to programs (e.g., Replicator, AUKUS Pillar I, Sentinel ICBM) and to platforms (“who builds the B‑21,” “who invented Lattice OS”), because authority and program prompts test precisely for builder credit.
↓
Everything PR34 / 39
● Vendor Comparison
How do you compare vendors effectively using AI without missing under-cited leaders?
Use AI answers for discovery but corroborate with primary-source and contract data (USAspending.gov, DoD announcements), since some legacy primes are under-cited while challengers are over-cited relative to revenue.
↓
Everything PR35 / 39
● PR Attribution
How can defense PR teams ensure their brand is credited accurately in AI answers?
Feed engines with structured primary sources that explicitly attribute inventions, platforms, and program roles; authority prompts like “who builds the B‑21” and “who invented Lattice OS” probe exactly those claims.
↓
Everything PR36 / 39
● Trade Press Role
How does trade press coverage affect whether a defense vendor shows up in AI answers?
Defense trade coverage in outlets like Defense News, Breaking Defense, Aviation Week, and Inside Defense still matters and is weighted by LLMs, but it isn’t sufficient on its own. The engines elevate vendors that also publish structured, entity-rich, primary-source materials.
↓
Everything PR37 / 39
● Earnings Influence
How do earnings calls and investor filings influence AI answer visibility for defense vendors?
Earnings-call narratives and investor filings operate as primary-source inputs that AI answer engines reward and weight. Alex Karp’s earnings-call performances exemplify how such communications elevate visibility. Palantir’s Q1 2026 10‑Q and Lockheed Martin’s 2025 10‑K provide structured, entity‑rich documentation that helps engines credit entities and validate claims.
↓
Everything PR38 / 39
● Presence-contract Gap
What does the gap between AI citations and defense contract dollars look like?
Defense-tech challengers like Anduril, Palantir, Shield AI, and Helsing are over-cited versus their revenues, while legacy primes (Lockheed Martin, Northrop Grumman, RTX, General Dynamics, Boeing Defense) are under-cited relative to market scale. Services integrators (Leidos, SAIC, CACI, Booz Allen Hamilton, L3Harris) are functionally invisible, and European primes punch below their post-Ukraine order books; Israeli firms are most under-cited versus combat records.
↓
Everything PR39 / 39
● Progress Measures
What should defense marketers measure in AI answers to track progress over time?
Track citation presence, ranking position, sentiment, source quality, and recency across ChatGPT, Claude, Gemini, Perplexity, and Google AI Overviews—and re-test to gauge visibility and citation stability over time.
