Ask an investor where a winning thesis came from, and they'll rarely start with numbers. Sure, they've got someone cranking on a model in the back, and they'll only make a decision if those numbers tie, but when they describe where their interest actually started, they talk about conversations. A former VP of sales who told them, almost as an aside, that renewals had gotten harder this year. A hospital procurement director who said she'd never heard of the company that was supposedly winning her segment.
The spreadsheets come later. Great investors, first and foremost, are great listeners.
Everyone Has the Same Data
The job of an investor isn't to say whether a company is good or bad; it's to determine whether that company is better or worse than the market's consensus view of it. The problem is that markets are smart, and they have access to the same information you do. Public filings, earnings transcripts, alternative data vendors. It doesn't matter how sharp your insight is — it's almost impossible to hold an edge when every sophisticated firm on the street is working from the exact same inputs.
Besides, those datasets are all missing something: they say nothing about what the people closest to a business actually think is happening. A company might report churn, but only the people watching closely can tell you whether that churn is a blip or a pattern. A company might launch a shiny new product, but only its target customers can tell you whether that product is actually moving deals. Investment models are built on assumptions. The information behind those assumptions lives inside people's heads, and the only way to get it out is to ask.
This is why expert interviews became standard practice at every serious investment firm, and why the expert network industry grew into a multi-billion dollar business without most people outside finance ever hearing about it. Primary research is table stakes. The diligence process runs on it — pre-IC calls pressure-test the thesis, channel checks validate demand, post-close conversations monitor what's changing.
What It Looks Like From the Investor's Side
You have a thesis and only three weeks. You start by writing a screening spec.
An expert network sends back a list of candidates, most of them adjacent to what you asked for rather than dead center. You pick five or six, wait several days for scheduling, and hope the two you actually wanted don't cancel. You take the calls yourself, or you hand them to whoever on the team has the lightest week.
You learn things. Real things. Sometimes they change how a model is wired; occasionally they kill a deal outright. But you also finish the process knowing that you sampled six people out of a population of hundreds, and that if you'd spoken to six different people you might have heard something different. You proceed anyway, because the deadline doesn't care.
Oh, and by the way: those six calls cost your firm 10 grand.
What It Looks Like From the Other Side of the Call
You're a twenty-year veteran of your industry. Someone emails asking for an hour of your time. You fill out a screener with eleven questions, half of which are legal boilerplate. You go back and forth on scheduling for a week. You get on the call, and it's a 24-year-old analyst reading questions off a page, typing furiously, cutting you off at minute fifty-two so they can prorate the honorarium. You never find out what they concluded. You wait five weeks to get paid.
Nobody designed this experience on purpose. It's just what happens when the administrative cost of an interview is high and the party absorbing that cost isn't the one who benefits from it. The expert bears the friction. The investor bears the delay. Neither is getting what they came for.
Oh, and by the way: most of that 10 grand went to cover the expert network's margin.
The Constraint Was Never the Experts
I've been that 24-year-old analyst. When I was a consultant, we didn't limit our primary research because there weren't enough people worth talking to. There were hundreds. We limited it because a human being can conduct maybe five good interviews in a day, and we had four analysts and three weeks. The ceiling on our research was our own calendar.
That constraint shapes everything downstream. If interviews are scarce, you spend them on the questions you're most confident matter. There's no time to chase down curiosity. You sample narrowly. You can't segment, and you certainly can't go back and re-interview the same cohort a quarter later to see what moved.
An entire product grew up around a scarcity that most people now mistake for the natural shape of the work.
What AI Actually Changes
AI is not going to tell you whether to do the deal (and if it does, please don't listen to it — you're a fiduciary, for chrissakes). It doesn't know which questions matter, and it has no idea what your fund's edge is. That judgment is yours, and it isn't going anywhere.
What AI is good at is the part in between — asking questions consistently, at scale, without getting tired at 4pm. An AI moderator runs the fortieth interview exactly the way it ran the first. It doesn't lead the witness because it's behind schedule. It doesn't skip the follow-up question because it's still typing the last answer. It transcribes perfectly, every time, so nobody has to choose between listening and taking notes.
For the investor, that turns a sample of six into a sample of sixty. And sixty isn't just six with more confidence; it's a different study. You can segment by geography, by company size, by whether they churned. You can run the same panel again in ninety days. You can chase the contradictory data point instead of noting it and moving on. Speed matters too: a study that took three weeks to schedule can run in days, which means primary research becomes something you do during diligence rather than something you wish you'd had time for.
For the expert, the change is more immediate. No scheduling ping-pong — take the call when you're free, including at 9pm on a Tuesday. No eleven-question screener written by someone who doesn't know your industry. No analyst rushing you off the phone. And no five-week wait to get paid. That one isn't even really an AI thing; in the age of APIs, there's no reason it should ever have taken that long.
Both sides have been absorbing the cost of the same missing infrastructure.
Where We Come In
Clairvoyant Research is building that infrastructure. We surface the right experts, conduct the interviews with an AI moderator that actually listens, pay participants in under a day, and hand back synthesized, compliant takeaways instead of forty hours of raw transcript. The judgment stays where it belongs — with the people making the investment decision.
If you'd rather run sixty interviews than six, or if you're an expert who'd like to be treated like one, we'd like to hear from you.