Run Your Pre-Call Screening as a Gate — and Get an Audit Trail for Free

Most expert-call screening runs on the honour system. Someone is supposed to ask the conflict and restricted-list questions before the substantive conversation begins. Nothing structurally stops the call from moving on if they don't. And weeks later, when someone asks "was this expert screened, and how?", the answer is a recollection rather than an artifact.

For research firms and investment teams whose entire premise is talking to people close to the companies they study, that gap is uncomfortable in proportion to how well the program works. The better your sourcing, the closer your experts sit to material information, and the more the screening step matters precisely when it is most likely to be rushed.

Screening as an afterthought

The failure mode here is rarely misconduct. It's drift.

Screening questions get compressed when a call starts late. They get skipped when the interviewer assumes the expert was already cleared upstream. They get asked after the expert has already started volunteering exactly the kind of detail the screen was meant to catch first. Each of these is individually defensible and collectively corrosive — because the control only works if it runs before the substantive conversation, every time, without depending on anyone's discipline in the moment.

And then there is the records problem. Even when screening is done properly, what you have afterward is usually a note that says it was done — not an independent trace of whether the conversation was actually allowed to proceed. If a question is ever raised, "we always ask those questions" is a process claim, not evidence about this call.

Make screening a gate, not a step

A more reliable model treats your screening as a gate rather than a step in a checklist.

You define the questions and the pass / hold / reject rules — your questionnaire, your tiers, your restricted list, your thresholds. The interview agent asks your screening questions first and only continues into the substantive interview if your own criteria are met. If they aren't, the call ends at screening. The substantive conversation cannot happen until the gate is cleared, because the part of the system that conducts the interview is not reachable until the screening criteria are satisfied.

The distinction matters. A "step" is something a well-intentioned person is supposed to do. A "gate" is something the conversation structurally cannot get past. The first depends on discipline under time pressure. The second does not.

A worked example

Consider a fund that maintains a conflict questionnaire: a set of yes/no questions, an auto-reject tier for hard prohibitions, and a hold-for-review tier for anything ambiguous. Today that questionnaire lives in a PDF and depends on whoever runs the call asking it first.

Run as a gate, the same questionnaire becomes the entry condition. An expert who answers cleanly proceeds into the interview on the same call. An expert who trips the hold tier — say, a current advisory relationship that needs a human to look at it — does not proceed; the call ends at screening and the matter goes to review. Nobody had to remember to stop the call. The questionnaire is still entirely the fund's; what changed is that satisfying it is now a precondition for the conversation rather than a courtesy at the start of it.

Consent and recording jurisdiction

The same shape covers a different obligation. Recording-consent and jurisdiction disclosures have to happen before substantive recording — and have to be demonstrable afterward. Run as a gate, the disclosure and explicit consent are captured first, and the interview only proceeds once they're on record. Decline, and the conversation ends there. It is the same mechanism as conflict screening, pointed at a different requirement.

The handoff is the audit trail

Here is the part that comes for free. Because the conversation only advances when your criteria are met, whether the handoff happened is itself the record. Cleared and proceeded, or ended at screening — that is observable after the fact, not reconstructed from someone's memory of a call months earlier. You don't have to instrument a separate logging step or trust a note; the structural fact that the interview occurred at all is the evidence that the gate was satisfied.

InsightAgent isn't a compliance product. You define your screening questions and rules; the interview agent follows them as a gate before the conversation proceeds, and the handoff is your audit trail.

What InsightAgent is — and is not — here

It is worth being explicit, because this is exactly the kind of capability that is easy to overclaim.

InsightAgent does not interpret regulations. It does not judge whether a relationship is a conflict. It makes no compliance or legal determination of any kind. It runs your screening logic — the questions and rules you define — and gives you an observable trail of whether that logic was satisfied on a given call. The judgement stays entirely with you and your compliance function; what the platform contributes is that the judgement is enforced structurally and is visible afterward.

InsightAgent maintains a SOC 2 Type II report covering security. That is a statement about how the platform is operated, not a claim that the platform is a compliance or legal product. The two should not be confused, and we are careful not to.

The net effect is narrow and useful: the screening you already designed stops being a checkbox that depends on a busy person remembering, and becomes a gate that the conversation cannot get past — with the proof of that built in rather than bolted on.

See how a screening gate is configured →