PRIMER AI

Sean Gourley did his PhD at Oxford studying the mathematics of modern warfare — specifically, whether insurgency violence followed predictable statistical patterns. The answer turned out to be yes.

He co-authored a paper in Nature in 2009 showing that the frequency of attacks in Iraq and Afghanistan followed a power law, the same distribution found in financial markets and earthquakes. That work attracted military attention.

After Oxford, Gourley co-founded Quid, a data visualization company, which he ran until 2015. When he left, he was thinking about the intelligence community's core problem: the US government generates and collects vastly more text data than any human team can read.

Every day, analysts at the CIA, NSA, and DIA are overwhelmed by reports, intercepts, academic papers, social media, and foreign news. Missing the important signal in that noise is not a hypothetical problem.

It has real consequences.

Gourley founded Primer in 2015 to build the machine reading layer between the raw information flood and the human analysts who need to act on it. The first investor was Lux Capital, which had backed deep tech and defense-adjacent companies.

The company's early focus was on natural language processing tools that could read, summarize, and extract meaning from unstructured text at scale — then bring what mattered to the surface.

Primer sells software to government agencies and defense contractors under multi-year contracts. Its primary product, Primer Command, is an intelligence analysis platform that ingests and processes text from multiple sources — classified and unclassified — and uses machine learning to identify relevant signals, patterns, and entities.

The contract model is typical of government software: slow sales cycles measured in months or years, rigorous security requirements, long contract terms once awarded, and significant switching costs. Primer has positioned itself as a trusted vendor with appropriate security clearances, which is not a moat that commercial AI companies can easily replicate.

Revenue is almost entirely recurring from government contracts. The company has not disclosed specific revenue figures, but reported contracts with the US Army and multiple intelligence agencies.

The 2021 Series C at a reported $525 million valuation suggested the investor base saw a clear path to $50-100M in ARR from the government market.

Primer Command — the flagship intelligence analysis platform for government and defense customers. Ingests structured and unstructured text from multiple classified and unclassified sources and surfaces entities, relationships, events, and patterns.

Used by US Army units and intelligence analysts for operational awareness and intelligence production. Primer Analyst — a tool focused on document analysis and summarization for individual analysts who need to process large volumes of text quickly.

Primer Copilot — an LLM-integrated interface launched post-2022 that lets analysts interact with their data through natural language queries, built to run inside secure environments where data cannot leave the network.

Primer's core growth strategy is expand-within-government: win one agency, prove value, and let the US government's inter-agency information sharing culture do the work. A contract with the US Army creates visibility at the Air Force and Navy.

A CIA relationship opens conversations at allied intelligence agencies. The security clearance infrastructure, once built, becomes a perpetual advantage.

Post-2022, as large language models became mainstream, Primer shifted its messaging toward integration — the idea that its platform gives government customers LLM-powered analysis inside secure environments where classified data cannot go to commercial APIs. This positioning is strong: classified data cannot be sent to OpenAI's servers.

Primer's on-premise and air-gapped deployment capability addresses a problem that no commercial AI product can solve.

International expansion follows US government relationships. If the US Army uses Primer, allied militaries take notice.

Five Eyes partners — UK, Canada, Australia, New Zealand — are natural expansion markets.

Competing with both established defense contractors (Palantir, Booz Allen Hamilton, SAIC) and new AI entrants (Scale AI, Anduril's AI subsidiaries) for the same government dollar. The government AI market is large but contested.

Primer has the advantage of being early and purpose-built for text intelligence. It has the disadvantage of being much smaller than Palantir and lacking the political relationships that larger primes have built over decades.

The LLM wave is genuinely double-edged. It validates everything Primer has been saying about machine reading since 2015.

It also means that every defense contractor now has an AI story and every government buyer is getting pitched AI-enhanced analysis from a dozen vendors simultaneously. Standing out in a crowded government AI market requires winning contracts and renewals, not just demos.

Lux Capital led the early rounds and has been a consistent investor across multiple financing cycles. In-Q-Tel — the CIA's venture arm, which also backed Palantir — participated in the Series B, providing both capital and the credibility signal that matters enormously in the defense AI market.

Felicis Ventures joined in Series B. Greenoaks Capital led the Series C in 2021, which raised $110 million at a reported $525 million valuation.

Greenoaks is known for large concentrated bets on software businesses they believe can become dominant in their categories — their Primer investment was a bet on the company owning the machine reading layer of US government intelligence.