AI in Marketing

What KYC-grade document scanning taught us about booth capture

Running events for an identity-document AI company means watching bank-grade capture standards meet show-floor reality. Five rules that transfer.

Prasad Subrahmanya avatar
Prasad Subrahmanya
Founder & CEO, Luminik, August 8, 2026, 3 min read

TL;DR: Five rules from bank-grade identity scanning transfer to the booth: the moment is unrepeatable, confidence is a first-class output, design for the worst venue, validate at the edge, and treat the scan as evidence rather than the asset.

Before Luminik, I spent years as an extended member of the marketing team at a leading identity-document AI company, the kind whose scanning runs inside banks’ onboarding flows. My job was their events: the booths, the target lists, the follow-up, the pipeline reporting. Which meant I stood at trade-show stands representing capture technology held to a regulator’s standard, watching our own badge scans fail in ways the product I was demoing would never be allowed to. The engineers’ rules were on one side of the booth and the event workflow’s reality was on the other, and the gap between them is a large part of why Luminik exists. Five of those rules transfer directly to scanning badges at a booth.

1. The capture moment is unrepeatable

In KYC, the user is in your flow once; if the scan fails, most never come back. On the show floor it is stricter: the prospect walks away in ninety seconds and there is no retry link to send. Everything about capture UX follows from this: one-handed operation, instant feedback on what was read, and a manual path that takes seconds when the automatic one struggles. Any design that assumes a second attempt is a design for lost pipeline.

2. Confidence is a first-class output

Document systems never say just “John Smith.” They say what was read and how sure they are, per field, because downstream decisions differ: a confident name auto-fills, an uncertain one gets human review. Booth capture deserves the same contract:

  • Confident fields write to the CRM untouched.
  • Uncertain fields surface for a one-tap fix while the person is still there.
  • Silent guessing is the one behavior that is never acceptable, because a wrong company name poisons enrichment, routing, and attribution downstream.

3. Design for the worst venue, not the demo room

Bank-grade scanning is tested against kitchen-table lighting, cracked screens, and ten-year-old phones. The show-floor equivalents are fixed and predictable:

KYC condition Show-floor twin Design consequence
Kitchen-table glare Expo hall downlights on laminate Read across multiple frames, not one shot
Old device cameras Reps’ personal phones On-device processing that degrades gracefully
No connectivity mid-flow Convention-center Wi-Fi Offline-first queue, sync when the network returns
Unfamiliar document layouts Every organizer’s new badge design Layout-agnostic reading, not per-show templates
Capture sits mid-pipeline. What happens before and after the scan decides its value.

4. Validation belongs at the edge

In identity flows, a birthdate that parses as next year is rejected at the point of capture, not discovered in an audit. Booth capture has the same class of cheap, decisive checks: a company field that reads as a country name, a title that is the event’s tagline, an email domain that does not exist. Every error caught at the booth costs one tap; the same error caught during CRM cleanup costs a meeting that never got booked.

5. The scan is evidence; the asset comes later

A passport image is worthless until it becomes a verified identity attached to an account. A badge scan is worthless until it becomes a routed record: ICP score attached, owner assigned, next step written, event attribution set. This is why capture-only tools plateau. The scan is evidence; the asset is the attributed CRM record, and producing it takes the stages around the scan. In the Series C cybersecurity program that ran this full motion across RSA and Black Hat, 1,840 ICP matches from 43,000 attendees became 85+ qualified meetings and $2.4M in attributed pipeline, and the badge scan was exactly one stage of that.

The transfer test

I did not build the scanning technology; I ran the event motion next to the people who did, and the standard rubbed off. If you evaluate capture tooling this season, borrow the KYC lens wholesale: does it treat the moment as unrepeatable, expose its own uncertainty, survive the worst venue, validate at the edge, and hand off into a pipeline rather than a pile? Two yeses is a scanner. Five is a stage of a system. The during-event capture page shows how Luminik answers the five, and where the stage hands off to attribution.

Prasad Subrahmanya avatar
About the author
Prasad Subrahmanya
Founder & CEO, Luminik

Founder of Luminik. Previously Venture CTO at Bain & Company and cofounder at Mainteny. Writes about how mid-market B2B teams build predictable pipeline from events.

Connect on LinkedIn

See how Luminik would approach your next event

A 20-minute walkthrough, tailored to the events on your calendar.