Clay Bootcamp is now Bootprint 🥾 The story →
Event Networking

From a conference badge to a live networking app

Vikrant Rao built a full conference networking app using Claude Code with zero coding experience. Point a phone at any event badge, get real-time ICP research and conversation starters in seconds. Weekend project. Now integrated into a product.

📅 August 12, 2026 ⏰ 62 minutes 🚀 All levels
▶️ Watch the Full Recording
Bootprint

Bootprint • Claude Code Cohort

Want to build tools like this for your own GTM motion?

One week, hands-on, small group. You go from zero to a working Claude Code setup with agents, skills, and MCPs your team can actually use. No coding background required.

Join the Claude Code Cohort Coached by GTM engineers like Vikrant • Bootprint

Guest Speaker

Meet Vikrant Rao

Vikrant Rao

Vikrant Rao

Co-Founder, RevenueGenie • Head of Marketing, Ottosales AI • GTM Coach, Bootprint

Vikrant has 8 years in B2B SaaS marketing and demand gen, with stints at Freshworks, LinkedIn, and SpotDraft. Two years ago he co-founded RevenueGenie to help B2B companies upgrade their GTM systems with AI, and recently joined Ottosales AI as Head of Marketing. Otto is an AI sales coordinator that lets field reps log meetings and send follow-ups through a phone call, no laptop required.

The networking app he showed today started as a weekend project the Friday before SaaStr. It now ships as a feature inside the Ottosales AI product. He built the entire thing with Claude Code and has never written a line of code.

Connect on LinkedIn

Session Overview

What we covered

📷

Badge scan to ICP match in seconds

Point a phone at any event badge. The app reads the name and company, triangulates the LinkedIn profile, and tells you whether it's a good fit before the conversation even starts.

🛠

Building without coding

Zero coding experience. Vikrant built the first version in 3 to 4 days using pure logic stacking and debugging in Claude Code. Architecture clarity matters more than knowing how to code.

🔌

Modular API architecture

Designed from day one to swap any API without rebuilding. RapidAPI, Apify, Tavily, Firecrawl, Gemini, SendGrid, and more. Each module is independent.

🎯

Free tool as GTM motion

Give the app away for free to field sales teams at events. High upfront value, zero barrier to respond, instant time to value. A better icebreaker than any cold pitch.

The Build

How the app works, step by step

1

Sign up with a company email

One agent researches your company domain in about 60 seconds and sets your ICP criteria automatically. You can adjust it any time in settings.

2

Open the app at the event and tap Scan

The camera opens. Point it at the person's event badge. Claude's native Vision OCR reads the name and company off the badge image.

3

App triangulates their LinkedIn profile URL

Simple Google search: name plus company name plus "LinkedIn profile URL." Checks the first 10 results. App confirms the match before proceeding so you're never looking at the wrong person.

4

Processing splits into two parallel lanes

Fast lane delivers ICP fit and company size match in seconds. Slow lane runs deep research in the background. The quick result lands while the full card is still loading.

5

Full card loads: research, insights, and conversation starters

Career arc, company overview, recent news, technographic signals via Theirstack, and a curiosity question tailored to what your company does. Profile photo included so you remember who you talked to.

6

Add a voice note and get a suggested follow-up

Record anything said in the conversation. The app generates a suggested follow-up message you can send on email or LinkedIn after the event.

7

Export all contacts as CSV at the end of the event

Every person scanned, with full research and ICP classification, exports in one CSV. Push it to your CRM manually or, soon, sync directly to Auto via Ottosales AI.

Frameworks & Mental Models

The thinking behind the build

The Good Offer Framework

Three pillars: High upfront value. Low barrier to respond. Fast time to value. Most audits fail all three. A free tool that works in 60 seconds passes all three. If someone says yes, they should get the benefit immediately.

Inspired by Jordan Crawford

Logic Stacking Over Coding

If the orchestration logic is clear in your head, Claude Code can build the thing. Knowing what you are building matters more than knowing how to code it. Start with a clear map: this API goes here, this response feeds this next step.

The Modular Architecture Rule

Tell Claude Code from the start: keep the code modular so you can swap any API with something else. Vikrant switched models mid-build when cost went out of hand. Two-line change per module because the architecture allowed it.

Parallel Processing for Real-Time UX

Find the fastest useful signal and surface it first. Run the slower, deeper research in parallel. A minute and a half wait kills engagement. A 5-second ICP match keeps the conversation going while the rest loads.

Cross-Question Before Fixing

When something breaks, ask Claude Code to tell you why it is breaking before asking it to fix anything. Without the diagnosis, it will start fixing the wrong thing and loop. "Tell me what is broken and why" first, every time.

The Smart Intern Mental Model

Claude Code knows everything about everything but needs to be pointed in the right direction. If you do not guide it, it will take you in loops. Your job is the logic. Its job is the implementation.

How Vikrant Prompted Claude Code

Instructions that shaped the build

These are the key instructions Vikrant gave Claude Code during the build and debugging sessions.

Architecture Instruction
Keep this entire code extremely modular. I should be able to switch any API that I want with something else.
Debugging Instruction
First tell me what is broken and why this is not working. Then we will fix it.
OCR Discovery
I want to scan event badges because that is the easiest way to use this at a conference. I need an OCR. How do we go about this?

Attendee Q&A

Questions Vikrant answered

QCan we do a bulk upload of the attendee list instead of scanning one at a time?
Not built yet, but straightforward to add. All the app needs is a name and company name to find the LinkedIn URL and run the research. A bulk upload input that feeds the same process for each row would work. The base logic already handles it one at a time, making it parallel is the only change.
QIs the flow to download as CSV and push into the CRM?
Yes, for now. You export all scanned contacts as a CSV at the end of the event, then import it to your CRM. They are building a direct sync to Auto inside Ottosales AI so field reps can skip the manual step entirely.
QWhat happens if it finds the wrong person?
If there are two matches, the app shows both options and asks you to confirm. If it finds nothing, it falls back to asking for a LinkedIn QR code scan. LinkedIn QR code is the universal source of truth. Once it has that, the rest of the process is the same.
QWhat is the ballpark cost to run a scan for one person?
Vikrant mapped each API call, what it costs per hit, and how many tokens the AI consumes to collate the data. That gives a base cost per scan. He adds a 20 percent buffer because token size can go up or down. The structure: API cost per call plus AI token cost plus 20 percent buffer equals your per-scan estimate.
QCan this connect to LinkedIn Sales Navigator so I know if someone is already on my list?
LinkedIn does not offer official API access to Sales Nav data or connection lists. The workaround: export your ABM list or CRM accounts, and build a check into the scan flow that queries whether the person exists in that list. If they are in your CRM already, it can pull the history too. One small module to add.
QWhich APIs do you recommend for pulling LinkedIn data?
Two main options: RapidAPI's Professional Network Data (covers profiles, posts, company info, job listings, and more) and Apify (has individual actors for different LinkedIn use cases). Heather also recommended the HarvestAPI LinkedIn Profile Scraper on Apify, which does not use cookies. For a one-stop GTM data provider, Deepline or AI Ark both run waterfall enrichment for emails and phones.
QDoes it enrich phone numbers and email addresses?
Not in this version. At events, business cards still get exchanged and the business card scanner captures contact details directly from the card. For digital enrichment, you could plug in Deepline or AI Ark. They run waterfall logic across multiple providers (Apollo, LeadMagic, Prospero, and others) to find phone and email.

Tools Used or Referenced

The stack

Claude CodeBuilder. Entire app built here.
Claude Vision OCRBadge and business card scanning
RapidAPIProfessional Network Data for LinkedIn
ApifyLinkedIn profile and company scrapers
HarvestAPILinkedIn Profile Scraper (no cookies)
TavilyWeb search and research
ExaSemantic web search (alternative)
FirecrawlCompany website scraping
TheirstackTechnographic data (tools on a site)
BuiltWithTechnographic data (alternative)
GeminiVoice note transcription
SendGridOTP authentication
SupabaseDatabase (free to start)
DigitalOceanHosting (~$3-4/month)
Next.jsFrontend framework
Tailwind CSSStyling
StripePayments (for public launch)
DeeplineEmail and phone waterfall enrichment
AI ArkEmail and phone waterfall (alternative)
GitHubCode hosting (free)

Resources

Links from the session

🎥
Session Recording
Full walkthrough with live demo (Zoom)
📷
Try the App: buddy.ottosales.ai
Free credits available. Scan a badge or QR code and see it live.
🔗
HarvestAPI LinkedIn Profile Scraper
Heather's recommended Apify actor for LinkedIn data. No cookies.
Bootprint

Build your own GTM tools with Claude Code

One week, hands-on, small group. You go from setup to a working agent your team can use. No coding background required.

Join the Claude Code Cohort

Stay in touch

Questions about Bootprint?

Vikrant Rao
Vikrant Rao
Co-Founder, RevenueGenie • Head of Marketing, Ottosales AI
Connect on LinkedIn
Heather Melton
Heather Melton
Head of Community Strategy, Bootprint • Founder, Loro Flow
Connect on LinkedIn