JobLink

The challenge: industrializing HR pre-screening with AI

Reduce recruiters' manual workload on shortlisting while ensuring an objective, repeatable assessment. Design a solution able to conduct phone interviews, check candidate-role fit and produce structured assessment reports.

A step-by-step approach: technical POC, V1 then V2

Three phases: a technical POC to validate the technology choices, a V1 of the platform deployed on their servers, and a V2 focused on WhatsApp integration and automated follow-ups.

1. Technical POC: validating the AI and voice choices

A prototype to test the solution over a short phone call and generate a report. Testing many prompt variants, exploring voice solutions, and a cost study estimating production expenses.

2. V1: platform and deployment

Feature specification, mockup creation, development and deployment on their servers. Successful adoption: more than 10,000 interviews scheduled and conducted by the AI agent.

3. V2: WhatsApp integration, follow-ups and candidate calls

Advanced WhatsApp integration, automated follow-ups by email, WhatsApp and SMS, and the ability for the agent to trigger phone calls to candidates.

See Smart Interview in action

Explore the AI pre-screening platform: automated interviews, scoring and assessment reports.

Tech stack: web, voice and AI

ElevenLabs for voice synthesis, Twilio and WhatsApp Business to schedule and conduct interviews. On the AI side: AI SDK, OpenAI and ElevenLabs. Frontend and backend: React, NestJS, PostgreSQL.

Front end: React, TypeScript, Vite, Tailwind CSS

Back end: NestJS, PostgreSQL, Redis

AI: OpenAI, ElevenLabs

Infrastructure: Docker, AWS, GitHub

Tools: Figma, PostHog

Results: adoption and impact

• POC validated: technology choices confirmed, with a cost study to guide the investment decision.
• V1 deployed: more than 10,000 interviews scheduled and conducted by the AI agent on the platform.
• Successful adoption: the platform is used in production and meets the pre-screening needs.

A continuous collaboration: iteration and adaptation

Iterating on prompt variants, testing the technical options for the voice engine, and adapting the product to every piece of user feedback.

Regular dialogue with the product and HR teams to incorporate field feedback, refine the assessment criteria and improve the candidate experience.