Every investor or consultant I speak to mentions some variant of how their firm is trying to use AI to do more in a challenging market. PE firms trying to source and screen more deals or trying to build more conviction and answer open questions more deeply about assets that get presented to Investment Committees without inventing more hours in the day.
My belief is that a new AI product in the primary research toolkit can help to solve for exactly these challenges; AI moderators that lead expert interviews. Whilst it seems like a recent development, behind the scenes, this shift has been in the works for quite some time.
At proSapient, my job is to uncover and bring our clients the latest tech and innovations for their primary research workflows, to save them time without compromising the quality of their work. We started exploring AI moderators over two years ago and really zeroed in on what makes a quality AI-moderated interview, their potential shortcomings and how we can work with leading tech providers to fit this product as smoothly into our clients’ day-to-day as we can.
As one can imagine, there is a long list of learnings, and I want to share some of these in this article. If you are a private equity professional and need to rethink your primary research spiel in the age of AI, then read on.
The first question everyone asks about AI-moderated interviews (AIMIs) is if they are actually good at conducting interviews. The second, almost immediately, is whether or not they should be worried…
The technology is improving by the month, but it is still no substitute for a skilled human on the other end of a call. An AI can follow a script, utilise market context and ask dynamic follow-ups, but it struggles in exactly the places where the best investment professionals and human interviewers leverage their domain expertise, navigating specialist context, pattern matching across decades of reference experience, chasing unexpected answers, and reading the subtle cues (hesitation, body language, manner of speaking), the signal that there is more beneath the surface. Our partners at Strella, Listen Labs and Outset, to which we supply high quality expert respondents, are making real progress on that last part, so it may just be a matter of time.
What AIMIs are good at is scale. They won’t replace the human interviewer, but they can extend the interviewer’s reach, lift response rates, and bring voices into your research that would otherwise never make it in.
These are the use cases where, in my experience, AIMIs show their real value.
The first and simplest use case is one that is not based on increasing response rates or decreasing headcount needs: Surveys.
Surveys are effective at collecting large numbers of quantitative insights from a wide audience, usually popular for consumer insights, product testing or post-investment monitoring. Where they start to struggle is with open-ended questions. Anything that is more anecdotal, qualitative or requires respondents to describe their feedback in their own words. That is where AIMIs comes in.
Respondents will jump on a call with an AI interviewer and answer the survey questions in a two-way conversation. This use case seems like the most straight forward win-win scenario for AIMIs. You get deeper, richer data points, since respondents will share more in a conversational format than simply being asked to write something down. The respondents have a better experience, as they are more engaged in a conversation than a box-ticking exercise. An experience that leaves both sides better off.
Besides these benefits, it is also the simplest to implement. The other use cases in this article require rethinking existing workflows, but in the case of surveys, the structure of running them already exists at most firms. The only change is methodology; AI-moderated interviews. Everything else in the workflow; defining your target audience, screening and vetting respondents thoroughly, structuring your questions, all stays the same. You just get better data and insights out at the end, so to me, this is an obvious choice.
Where I see the first signs of rethinking current workflows, is when AIMIs are used within the broader context of due diligence workflows of PE deal teams, but only if teams are willing to adjust the workflow around them.
Let’s start by looking at the traditional workflow. Typically, a deal team will identify an opportunity in the market. They will speak to a few experts, typically an ex-C-suite operator or two, to build a point of view on the asset and its market. This work culminates in an internal go/no-go decision after which a formal commercial diligence kicks-off at much larger scale. The purpose of this initial screening is to understand if an asset is worth continued investigation and by extension if a larger-in-scope commercial diligence is worth it. This means that during deal screening, time and capital are constrained, while deeper, better-quality insights at this stage reduce the risk of wasting an Investment Committee’s time or, worse, making the wrong decision at go/no-go.
Therefore, the stakes are high and the pressure is on. This is exactly where AI can act as an accelerator, building conviction whilst keeping teams lean and research spend in check. But let’s see how this could play out in action.
Imagine you are looking at a deal in a market you don’t know well. If you’re following the traditional deal screening model, you might be talking to two or three, maybe even five experts that help shape your view. Hearing from five great experts takes time and only gives you so many perspectives, the only way to have an educated view on the market is, consequentially, more insights.
If you are adding AIMIs into the mix, you might still want to do those crucial first calls with experts yourself, to help get smart in an engaging way. Then, however, you let AI take over and interview ten more experts (or even twenty, if you fancy). While you are building your understanding on a live call with the highest-stakes experts, AI has already delivered additional transcripts to your inbox, helping you validate and confirm (or reject) your hypothesis. You can then go into your next Investment Committee meeting with the confidence of dozens of perspectives, rather than a couple.
Now imagine you are past deal screening, and you have a good understanding of the asset and market. What you are missing is the customer perspective.
If you haven’t opted for a survey, customer interviews can be some of the structured yet most time-intensive components of commercial diligence, which means it is the perfect use case for AIMIs. The goal again is not to kick-out human judgement, but to scale where it currently creates bottlenecks or high costs. A good way of running such a project, in my opinion, is to let AI handle large amounts of data collection in interviews, since you can get qualitative and quantitative data at the same time, then, once the interviews are ongoing and the results come trickling in, a quick analysis will usually surface some outliers. Those are the ones to focus on for live calls. Schedule follow-ups with outliers to dig deeper and get the nuance that only you can dig out.
Both of these examples require rethinking of the current workflow, but the outcomes and savings more than justify the effort.
Primary research offers the highest ROI of any diligence activity, converting ground-level insights into investment alpha. The marginal value of an accurate view during deal screening or in a competitive deal process is massive. That is why I believe AIMIs are going to be adopted by the buy-side and their advisors faster than in most other professional contexts.
Innovation in AI technology will only accelerate from now on too. Starting to build AI into existing workflows to make better investment decisions needs to start now, if you don’t want to fall behind competitors. Diligence processes will become faster but also shaped with richer data insights. That also means that high-quality, primary insights that are not readily accessible on the internet, will become even more important. Partnering with the right expert network will be crucial for successful deals, more so than ever.
proSapient is working hard and is the best at delivering trusted, high quality expert insights at speed, though I may be a little biased.