✦
The promise is real. The execution isn't there yet. AI-led expert calls are the hottest development in primary research right now. The pitch is compelling: an AI interviewer conducts expert conversations on your behalf, transcribes them instantly, and feeds the insights directly into your research workflow. No scheduling headaches. No analyst time spent on calls. Just scalable, automated intelligence.
We've tested them. We wanted them to work. But after putting AI-led calls through real investment research scenarios — the kind our clients run every day across PE due diligence and public equity research — we have to be honest about what we found: the quality isn't there yet.
This guide breaks down what's actually happening with AI-led expert calls, where the technology genuinely adds value today, and where human-led research remains irreplaceable for investment teams making high-stakes decisions.
What Are AI-Led Expert Calls?
AI-led expert calls are exactly what they sound like: instead of a human analyst or investor conducting a phone interview with an industry expert, an AI system runs the conversation.
The most prominent example is AlphaSense's AI Interviewer, which the company launched in August 2025. AlphaSense launched its AI agent interviewer and debuted Channel Checks, capabilities that expand Tegus Expert Insights, the company's comprehensive expert research offering. The product is designed to scale their existing transcript library — the AI Interviewer is already contributing hundreds of calls to the Tegus Expert Transcript Library each week, and the library will continue to expand beyond its current 220,000+ expert transcripts.
Here's how it works in practice: customers can launch AI-led expert calls in hours by selecting a topic, approving an AI-generated question guide, and receiving a fully transcribed interview with key takeaways. The process follows a structured workflow. Experts are sourced through AlphaSense's global expert network, pre-vetted for relevance and compliance. The AI conducts the live conversation by phone. Every call is recorded, transcribed, and reviewed for compliance before being published. Once published, transcripts are discoverable through the content library.
The vendor messaging is ambitious. AlphaSense claims the AI Interviewer "leverages the full breadth of prior relevant knowledge to engage senior industry experts in probing conversations, matching the precision of an experienced human interviewer while operating at the speed and scale only AI can deliver."
That's a big claim. Let's stress-test it.
Who's Trying AI-Led Expert Calls — and Why
The interest is real and widespread. The expert network industry is a $3 billion market growing at roughly 12% annually, and AI is the primary driver of new product development across the space.
Several forces are driving adoption.
Cost pressure on research budgets. Traditional expert network calls cost anywhere from $500 to $1,000+ per call depending on the expert's seniority. AlphaSense offers a flat rate of $400 per expert call, while other market intelligence platforms operate on a variable cost model where costs depend on the perceived quality of experts, ranging from $500 to $1,000 per expert. AI-led calls aim to push that cost down further by removing the human interviewer entirely.
The scaling problem. PE deal teams running commercial due diligence might need 20 to 50 expert conversations per project. Hedge fund analysts covering multiple names need ongoing channel checks across sectors. The bottleneck has always been analyst time — there are only so many calls a person can run in a day.
Speed to insight. In competitive deal processes and fast-moving public markets, the team that gets to a differentiated view first wins. AlphaSense positions its Channel Checks as enabling users to get a real-time pulse on the economy, surfacing market-moving insights such as demand shifts, pricing changes, and supply chain disruptions.
Consolidation is creating integrated platforms. AlphaSense announced the launch of Financial Data during its inaugural AlphaSummit customer conference, seamlessly combining in one chat interface structured quantitative financial data with proprietary qualitative insights like broker research, expert calls, and company filings. The vision is a single platform where AI-generated expert content sits alongside filings, earnings, and financial data — all searchable and cross-referenced.
The appeal is obvious. If it worked as advertised, every investment team would adopt it tomorrow.
What AI-Led Calls Actually Do Well
Let's give credit where it's due. AI-led expert calls do solve some real problems.
Standardised data collection at scale. When you need the same five questions asked to 30 distributors in a channel check, an AI can execute that consistently. It won't get tired, won't vary its phrasing, and won't accidentally lead the witness. Conducting channel checks once took weeks of manual calls and analysis. With the AI Interviewer agent, AlphaSense now conducts channel checks at scale, delivering real-time insights on supply and demand, pricing and volumes, and competitive dynamics across dozens of high-value industries.
Speed of execution. An AI can run multiple calls simultaneously, around the clock. For time-sensitive situations — a deal that needs to close this week, an earnings catalyst approaching — the raw throughput matters.
Transcript availability. AI-led calls produce instant, searchable transcripts. AlphaSense has enabled near-immediate access to raw, unreviewed expert call transcripts, available within minutes after calls. No waiting days for a transcription service. No arguing with your expert network about turnaround times.
Compliance consistency. Experts sign strict agreements prohibiting the disclosure of material non-public information. The AI Interviewer is tuned to avoid risky or inappropriate topics, and a dedicated compliance team reviews every transcript, removing or redacting flagged content before publication. An AI won't accidentally stray into MNPI territory mid-conversation the way a less experienced analyst might.
These are legitimate advantages. For certain use cases — particularly high-volume, structured channel checks where you need directional data points rather than nuanced insight — AI-led calls can be a useful addition to your research toolkit.
Forget feature matrices. When you’re evaluating a primary research provider, you’re evaluating seven things. Each one reveals something structural about how the provider operates and whether their model is built to serve your outcomes or their margins.
Where AI-Led Calls Fall Short: What We Found
Here's where it gets real. We've tested AI-led expert calls across the types of research scenarios our clients bring us every week — PE commercial due diligence, competitive positioning analysis, customer reference checks, market sizing exercises. The gaps are significant.
AI can't read between the lines. The most valuable insights from expert calls rarely come from direct answers to direct questions. They come from what the expert doesn't say. From the hesitation before they answer. From the offhand comment about a competitor that reveals more than any structured question could.
A skilled human interviewer picks up on these signals instantly and knows how to probe further. "You paused there — is there something about that relationship you're not sure you can share?" Or: "You mentioned their pricing 'used to be' competitive — what changed?"
AI doesn't read tone, hesitation, or subtext. It processes words. And in primary research, what matters most is often everything around the words.
AI can't build rapport or encourage openness. Industry experts — especially senior ones — don't just answer questions. They decide how much to share based on who's asking, how the conversation feels, and whether they trust the person on the other end of the line.
A good researcher builds rapport in the first two minutes of a call. They establish credibility, demonstrate that they've done their homework, and create a conversational dynamic where the expert feels like a peer, not a subject being interrogated.
AI can't do this. It asks questions. The expert answers them. There's no warmth, no reciprocity, no "that's a really interesting point — we've heard something similar from another perspective, which makes me wonder..." The result is that experts give shorter, more guarded answers. You get the Wikipedia version of their knowledge, not the real version.
AI follows scripts — it doesn't follow threads. The best expert calls aren't linear. A question about customer retention might surface an unexpected insight about a competitor's sales tactics, which leads to a revelation about market pricing dynamics that nobody on the deal team had even considered.
Human interviewers follow these threads in real time. They understand what's important to the specific investment thesis and can pivot instantly when something unexpected and valuable emerges.
AI interviewers operate from a question guide. Even with adaptive follow-up capabilities, they lack the contextual understanding of what matters for your specific deal or your specific thesis. They can't distinguish between a mildly interesting tangent and a thesis-changing insight that demands 15 minutes of deep probing.
Transcription and comprehension errors compound. AI-led calls introduce errors at multiple points. The AI may mishear or misinterpret what the expert says, especially with industry-specific jargon, accented speech, or technical terminology. These errors flow straight into the transcript that your team then relies on.
When a human leads the call, they can clarify in real time. "Just to make sure I understood — you said the contract renewal rate dropped to 70%? Was that across the whole portfolio or just the mid-market segment?" AI misses these moments, and the result is transcripts with subtle but potentially material errors that analysts may not catch.
The insight-per-call ratio drops dramatically. This is the bottom line. When we compared the actionable, thesis-relevant insights generated from AI-led calls versus well-structured, human-led calls on the same topics, the human-led calls consistently delivered more insight per conversation — often significantly more.
AI-led calls tend to produce surface-level, confirmatory information. Human-led calls uncover the surprises, the contradictions, the "here's what nobody's talking about" moments that actually move the needle on an investment decision.
Where AI Helps Today: The Right Use Cases
We're not anti-AI. We use AI tools extensively in our own research process. But we use them where they actually work, not where the marketing says they should.
Designing questionnaires and interview guides. This is where AI shines. Give an LLM your investment thesis, the target company, and the type of expert you're interviewing, and it can generate a solid first-draft discussion guide in minutes. It'll identify question areas you might not have considered, suggest probing follow-ups, and help you structure the conversation flow logically. A human researcher then refines and pressure-tests the guide before it's used, but AI accelerates the starting point significantly.
Summarising and analysing transcripts. After the calls are done, AI is excellent at processing the output. It can summarise key themes across 20 transcripts, flag contradictions between experts, extract specific data points into structured tables, and highlight areas where expert consensus is strong or weak.
Searching and cross-referencing existing transcript libraries. If you have access to large transcript databases, AI-powered search tools are genuinely useful for finding relevant prior conversations, identifying which experts have the most relevant experience for your current project, and triangulating what you're hearing against what others have heard before.
Generating first-pass research frameworks. For early-stage scoping — "what are the key questions we should be asking about this market?" — AI tools can synthesise publicly available information and suggest research frameworks that a deal team can then refine based on their specific investment lens.
The pattern is clear: AI is best as a force multiplier for human researchers, not as a replacement for them. It excels at the work that happens before and after the expert conversation. The conversation itself — the real-time, adaptive, human-to-human exchange where insight is actually created — is where AI falls short.
A Practical Framework: When to Use What
Not every research question requires the same approach. Here's how to think about matching your method to your need.
For high-volume channel checks — the same five questions asked to 30 or more contacts — AI-led calls can work. It's structured, repeatable, directional data collection.
For exploratory expert interviews where you're understanding a new market or thesis, use human-led research. Always. This requires adaptive questioning, rapport, and the ability to follow unexpected threads.
For PE commercial due diligence — customer references, competitive positioning — human-led research is essential. Experts share more with skilled humans, and nuance and credibility assessment are critical.
For thesis validation calls where you're testing specific assumptions, human-led is strongly preferred. Challenging assumptions requires adversarial, trust-based conversation.
For ongoing sector monitoring — tracking demand signals quarterly — AI-led calls can supplement. They're good for maintaining a baseline, with human-led calls for deep dives when signals shift.
For post-call synthesis and analysis, use AI tools with human oversight. AI excels at summarising, comparing, and structuring insights across multiple calls.
For interview guide and questionnaire design, use AI-assisted drafting refined by humans. AI generates strong first drafts. Human expertise sharpens them for the specific context.
What This Means for Your Team
If you're a PE associate running commercial due diligence, a hedge fund analyst building a thesis, or a corporate strategy professional evaluating an acquisition — the question isn't whether to use AI or humans. It's where each adds value in your specific workflow.
Here's our honest assessment.
Use AI tools aggressively for research preparation — questionnaire design, background synthesis, framework generation — and post-research analysis, including transcript summarisation, cross-call synthesis, and pattern identification.
Don't rely on AI to lead the conversations that matter most. The expert calls that actually change investment decisions — the ones where an expert reveals that the target company's biggest customer is actively evaluating competitors, or that the market is half the size everyone thinks — happen because a skilled researcher knew exactly how to create the conditions for that kind of candour. AI isn't there yet.
Consider a done-for-you model if your team is stretched thin. The real bottleneck for most investment teams isn't access to experts — it's the analyst time required to design the research, run the calls, and synthesise the output into something an investment committee can act on.
The Bottom Line
AI-led expert calls are an important development in the research landscape, and they will get better over time. The technology is improving rapidly, and investment from platforms like AlphaSense is massive — the company is investing over $100 million annually in the rapidly growing Tegus Expert Transcript Library, with a global team of hundreds of expert recruiters across all major industries.
But for now, AI's ability to lead conversations with humans, encourage them to open up, and extract information in real time while understanding what's important to a specific investment decision — it's not there yet. It makes too many mistakes, misses important nuance, mis-transcribes or misunderstands, and fails to encourage the kind of open dialogue that produces genuine insight.
The smartest investment teams will use AI where it works — preparation, analysis, synthesis — and keep humans where they matter most: in the room, or on the call, with the expert, doing the work that actually produces differentiated insight.
That's what we do at Woozle every day. Our clients brief us on what they need to know, and we handle the entire research process — designing the right questions, selecting and interviewing the right experts, and delivering finished, actionable research outputs. No scheduling calls yourself. No synthesising 15 transcripts. No hoping an AI didn't miss the most important thing the expert said.
If your team is evaluating how to get more from primary research — whether you're looking at AI tools, expert networks, or a different model entirely — we'd welcome the conversation.
This guide is published by Woozle Research for informational purposes. It is intended to help investment teams evaluate primary research providers and does not constitute financial, legal, or investment advice.
