In mature categories, AI engines are roughly accurate.
In frontier categories — quantum, post-quantum cryptography, Web3 infrastructure, AI tooling, novel-mechanism biotech — they hallucinate at a rate that should embarrass any communications team responsible for the founder.
I've now watched it happen across portfolios. A quantum-computing CEO who has never raised from a particular fund being credited with the relationship in ChatGPT. A DeFi protocol founder attributed to a different protocol with a similar name. A cybersecurity company's category miscategorized into adjacent-but-distinct enterprise software, costing it the right comparable set in every analyst spreadsheet built from an AI synthesis. A biotech compound described as treating a disease it has never been tested for.
These aren't edge cases. In emerging tech, this is the median outcome.
Three reasons it's worse at the frontier:
Thin authoritative corpus. In a mature category — payments, SaaS, pharma — the engines have decades of analyst reports, SEC filings, peer-reviewed research, and trade-press coverage to triangulate from. In quantum or Web3, the authoritative layer is thin. The training set is dominated by founder blog posts, Discord transcripts, Medium articles, and Reddit threads. The engines weight what's available.
Contested terminology. "AI agent" means six things. "Quantum-safe" means three. "Layer 2" depends on whose framework you accept. When the engines summarize, they collapse the distinctions and pick one — usually the one the most-cited source uses, which is often not the one the founder uses.
Founder-name collisions and category bleed. Frontier categories have small founder pools and high name-recall overlap. When two founders share a first name, a category, and a coast, the engines blend them. I have seen this happen at the CEO level, in writing, in an answer an analyst then forwarded to a partner.
The cost is not theoretical. It shows up as:
Investor decks built off the wrong comparable set
Talent that turns down the offer because the public summary made the company sound smaller than it is
Acquirers who never reach out because the engine put the company in the wrong bucket
A founder who has to spend the first ten minutes of every meeting correcting the record
The fix — same four moves, harder execution in emerging sectors:
1. Define the category yourself, in writing, on a page you own.
A canonical category-definition page with named entities, distinct from competitor categories, with primary-source citations. The engines will index it. They are starved for authoritative definitions in frontier sectors.
2. Disambiguate aggressively.
If your name, category, or technology overlaps with another entity, build the disambiguation into your own canonical sources. Wikipedia disambiguation pages exist for a reason. So should yours.
3. Publish technical primary sources.
Whitepapers, founder-authored category essays, structured FAQs. The engines reward technical depth in declarative prose. They cannot easily summarize a vague marketing page; they will happily summarize a 3,000-word technical explainer.
4. Audit on a thirty-day cadence.
Frontier categories move. Training-data refreshes move. A new competitor with a near-identical name will surface and the engines will collapse you into them within weeks.
For the CMO of a frontier-tech company, this is now a P&L line, not a marketing nicety. The cost of a misread is measured in slipped fundraises, missed acquisitions, and analyst notes that anchor to the wrong thesis for the life of the company.
The traditional answer was: get a Forbes hit, get a Bloomberg mention, let it cascade.
That still works. It also is no longer enough.
In emerging markets, the company that defines the category to the AI engines is the company that wins the category. The rest spend their next round explaining what they actually do.
Kyle Porter is Executive Vice President and Managing Director of Virgo Public Relations and a contributor to Everything-PR. He leads communications for emerging-industry companies across quantum computing, Web3, cybersecurity, biotech, and AI, including Quantum Art ($140M+ raise), OpenSea, Genies, and CleanStart.
