Diligence Engagements — AI-Led Interviews

The interview discipline of a research firm, at the speed and scale of software.

Heron's AI-led interviews reach the former employees, customers, and competitors who know a business — dozens at a time, in hours — using discussion guides and follow-up patterns refined across more than 50,000 researcher-led interviews. Same sourcing standards. Same compliance. A different lane for a different kind of question.

2,000+

AI-led interviews completed

50,000+

interviews in the Heron library informing every guide

48 hrs

first transcripts delivered

20+ yrs

of interview craft behind the discussion guides

One engagement, two lanes

Choose the lane per question, not the vendor per project.

Every Heron business diligence engagement can draw on both lanes. Researcher-led interviews go where the source is hard to reach and the nuance is the point. AI-led interviews go where the question needs breadth — many voices, fast, structured the same way every time. Most engagements use both.

Researcher-led
AI-led
Best for
Researcher-led
Hard-to-source people; senior former executives; nuanced or sensitive situations
AI-led
Breadth and speed; customer, channel, and competitor checks; post-earnings reads
Typical scale
Researcher-led
5–15 interviews per engagement
AI-led
20–100+ interviews per program
Turnaround
Researcher-led
2–6 weeks
AI-led
Hours to days
Who conducts
Researcher-led
A Heron researcher, trained on MNPI protocols
AI-led
Heron's AI interviewer, on researcher-built guides, with researcher oversight
Sourcing
Researcher-led
Heron's global sourcing team
AI-led
Same team, same standards
Output
Researcher-led
Transcripts, synthesis, decision-grade deliverables
AI-led
Transcripts, structured evidence, synthesis on request
Escalation
Researcher-led
AI-led
Any interview can be handed to a researcher
How it works

From question to transcripts in four steps.

01

Scope

You bring the question. A Heron researcher turns it into a discussion guide: the hypotheses to test, the follow-ups that matter, the lines not to cross.

02

Source

Heron's sourcing team identifies and screens former employees, customers, competitors, and channel participants of the company in question. No consumer panels.

03

Interview

The AI interviewer conducts each conversation on the guide, probing on what the source actually says, at whatever hour suits the source, in their language.

04

Deliver

Transcripts land in the Analytics Platform and the Heron MCP as they complete, with structured evidence and, on request, a researcher's synthesis.

The interviewer

Trained on how Heron researchers actually ask.

Most AI interviewers learned from consumer surveys. Heron's learned from twenty years of diligence conversations: when to ask a source to be specific, when to ask for a number, when to test a claim against what another source said. Every guide is built by a researcher; every interview follows it; every follow-up is grounded in what the source just said, not a script.

“The new pricing didn't land well with our larger accounts.”

ClarifyWhich accounts, roughly by size, pushed back hardest?

QuantifyOf those, how many did you actually lose or downgrade?

ContrastYou mentioned the prior year's increase went through. What was different this time?

Adaptive follow-ups

Probes on the source's own words. “Tense” becomes “what specifically did they point to.”

Consistent structure

The same guide across every source, so twenty interviews are comparable, not twenty anecdotes.

Researcher oversight

Guides are researcher-built; transcripts are reviewed; any conversation can be escalated to a researcher-led interview.

Use cases

Where breadth beats depth.

Customer checks at scale

Twenty to a hundred customers of a target on retention, pricing, and switching intent — before the LOI.

Channel and competitor mapping

Distributors, resellers, and rivals on where the pressure is coming from.

Post-earnings reads

Former senior employees within 48 hours of the print. This is how Heron Events runs.

See Heron Events →

Portfolio monitoring

Recurring quarterly programs on the same guide, so change is measurable.

Thesis screening

Ten quick interviews to decide whether a name deserves a researcher-led engagement.

Multi-market coverage

Sources interviewed in their own language, across the 20+ countries where Heron operates.

From an AI-led interview

“Nobody from the buyer side had ever asked me how many of my accounts saw the competitor's bundle. I'd never counted. It was eight of twenty.”

Former Regional Sales Director, enterprise software target · AI-led interview, 38 minutes
Compliance

The same rules the researchers follow. Enforced in software.

MNPI protocols

The guide encodes the lines a Heron researcher would not cross. The interviewer does not cross them either, and transcripts are reviewed before release.

Consent and recording

Every source is informed they are speaking with an AI interviewer on behalf of Heron and consents to recording before the conversation begins.

Conservative attribution

Sources are described by role and tenure, never by name. Heron's attribution framework errs toward less identifiability, not more.

Delivery

Transcripts where your team already works.

Completed interviews appear in the Heron Analytics Platform and through the Heron MCP as they finish. Integration partners carry the library to the Bloomberg Terminal, FactSet, Capital IQ, and Daloopa. Heron Skills turn a program's transcripts into your firm's own memo format inside your own AI.

When the AI reaches its limit

A researcher is one step away.

Some sources won't talk to software. Some conversations turn out to need a person who can read the room. When that happens, the interview moves to a Heron researcher — same engagement, same guide, no re-scoping. The two lanes exist so you never have to choose between coverage and depth.

Questions investors ask

The honest answers.

Will senior former executives really speak candidly to an AI?

Some prefer it: no scheduling friction, no interviewer to impress, and the conversation happens when they choose. For sources who won't, a researcher conducts the interview.

Does the interviewer just run a script?

No. The guide sets the ground to cover; the follow-ups are generated from what the source actually says. See “The interviewer” above.

How are sources found?

By Heron's sourcing team, against the specific company — former employees, customers, competitors, channel partners. Heron does not use consumer panels.

How is quality controlled?

Researcher-built guides, transcript review before release, and interview-quality grading on every conversation. Weak interviews are re-run or escalated.

Is it compliant for a regulated investor?

Yes. The MNPI protocols, consent standards, and attribution framework are the ones Heron has run for asset managers for more than twenty years.

How does it relate to Heron Events?

Heron Events is a scheduled application of AI-led interviews to post-earnings reads on covered companies. This page describes commissioned AI-led interview programs on the company of your choice.

AI-Led Interviews

Breadth in hours. Depth when you need it.