The starting point
Click to WhatsApp ads brought candidates straight into chat. That meant many conversations, but also many people who did not fit the job, and recruiters answered the same questions every day. Nobody knew which ad a candidate came from.
What I built
- Questions with buttons. WhatsApp lists and buttons, so candidates answer with one tap. Replies in other languages are translated by a language model.
- The ad source on every lead. The ad ID from Meta is turned into the campaign, ad set and ad name.
- Screening. Candidates who do not fit go to a separate pipeline, so recruiters only see the right ones.
- A reminder. People who stop halfway get a short nudge.
- Own WhatsApp to CRM gateway. It replaced a paid service at 90 USD a month and handles photos, PDFs and voice messages.
- Tested before launch. 9 of 9 tests from first message to CRM lead passed.
The result
From June to September 2026 the bot handled 526 conversations. 195 candidates answered everything and landed in the CRM with their ad source, 212 were screened out without a recruiter spending time on them, and 8 who stopped were won back by the reminder. Recruiters now call candidates who have already answered the key questions.
Tools and tech stack
- WhatsApp Business API
- Node.js
- Express
- SQLite
- OpenAI
- CRM API
- Meta Ads API
- nginx
The client is anonymised. The numbers come from the project's own reports and measurements.
Want this for your business? See AI automation services or WhatsApp chatbot for business.







