Program offices, congressional staff, and investors now begin their research in a chat window. We measure how AI systems find, resolve, and describe defense companies. Then we run the publishing engine that changes the answer.
Forty-seven seconds inside the console: share of answer by engine, the questions that matter in your market, the sources each engine actually trusts, and the ranked actions that move the number.
The discipline of a program office, applied to your public record. Instrument first. Then intervene, then measure again.
A full index run against your named competitive set: share of answer, featured versus passing placement, sentiment, and a map of every source the engines cite when they describe you. Branded and discovery queries are scored separately. Blending them flatters the companies that need the most help.
Where the answer comes from, and why. Entity failures: DBAs, CAGE codes, and legal names the model cannot connect. Hallucinated facts. Stale or hostile citations. And the open slots, the high-value questions with no incumbent answer that you should own.
A publishing and earned-media engine tuned to the retrieval pipeline: citation-ready capability pages, structured data shipped on your site, announcement timing discipline, and placement in the sources engines actually cite. We measure again every quarter and show the movement.
We publish weekly AI Visibility Indices for defense sectors, openly. The same instrument powers every client engagement.
We do not guess how models see the defense industrial base. We measure it at scale. Sector rosters. Twenty buyer questions per run, asked weekly with web search on, on the surfaces decision-makers actually use.
Every answer is scored for who was named, who was recommended, and which sources the engine trusted. The result is a share-of-answer index that shows who owns each question in your market, and which questions no one owns yet.
Six metrics built for how defense actually shows up inside AI answers, and how it fails to.
The share of scored discovery answers that name you. A plain mention rate: every query and every engine counts equally. Where the name first appears is tracked separately and does not change the percentage.
How much of your AI narrative rests on independent journalism versus your own website and paid directories.
Whether engines can connect your legal names, DBAs, CAGE codes, awards, and coverage to one company.
The credibility ladder of every domain the engines cite about you, from wire-service journalism to promo blogs.
High-value questions with no incumbent answer, ranked by how winnable they are for your capabilities.
Hallucinations and misattributions about you and your programs, caught before they propagate into derivative answers.
The tracker queries ChatGPT, Claude, and Gemini with web search on. The free check is narrower: ten buyer questions on Gemini, and the first three of those on ChatGPT and Claude as well. The public drone index currently runs Claude and Gemini, and each edition names the engines it used. Annual billing is 2 months free. This site does not take payment.
The tracker on ChatGPT, Claude, and Gemini.
The tracker, plus 6 customer-approved articles a month and a monthly readout.
The same three engines, scoped with you. No public price.
A one-time audit. The fee is credited toward the first month of a plan.
Request the Baseline →Ninety days, paid up front. No monthly invoice during the pilot.
Start the Pilot →We publish our sector research openly. Full datasets and briefings go to Enterprise subscribers.
A ~$100M private defense manufacturer won a major Army munitions prototype OTA, its best earned coverage ever. Seven outlets reproduced a shortened name that no AI engine could resolve to the company. Entity Resolution Rate on the biggest news in its history: zero. A competitor announced first and owns the story.
In the October 12, 2026 edition of the US Drone Industry AI Visibility Index, the four most-visible companies hold 46.5% of every mention across 40 answers from Claude and Gemini. The bottom 2 of 15 roster companies sit below 10%. The models rank media footprints. The people reading those answers assume they rank capability.
A hypersonics startup ran our free check twice. On three generic "defense sector" prompts, no engine named it once. On ten prompts written the way its actual buyers ask, every one of sixteen answers did. Visibility is not a property of a company. It is a property of a company and a question, and the generic question is the one nobody in a program office types.
We make true things visible. We never make false things plausible.
We onboard a small cohort of companies each quarter and review every application individually. Tell us who you are and we will be in touch about a Visibility Baseline for your competitive set.
Reporting on Congress and AI tools. These articles are not about Archimedes.