Problem & existing process
CVs Did Not Show Who Could Actually Sell on the Phone.
Before the platform, applicants sent CVs and the hiring manager screened them by hand, then called people one by one to see how they handled a conversation.
- A CV says little about how someone handles an objection on a cold call.
- Every first-round call took the manager's time, including for candidates who were clearly not ready.
- There was no common yardstick, so two candidates were never judged on the same questions.
- Results lived in chats and notes, with no history when the same person applied again.
Solution & what we built
A Mobile Application, a Live AI Call and a Report the Manager Can Trust.
The candidate completes a short mobile wizard, then makes a timed cold call to an AI business owner. The call is analysed against the hiring manager's own questions, and the admin panel shows a full report.
Mobile-first application wizard
Profile, availability, languages, experience, visa status, expected calls and meetings, and a required CV, in English or Arabic, with progress saved if the page is refreshed.
Live AI role-play call
A five-minute call with an AI business owner. Each attempt gets a different scenario (car rental, restaurant, dental clinic, furniture, beauty salon, contracting), and in some the person who answers is not the decision-maker.
Audio that works on local networks
Call audio runs through our own server over the standard web port, because some networks restrict direct browser voice connections. The server also keeps the transcript, so it cannot be edited by the candidate.
Scoring against the manager's questions
Ten questions written by the hiring manager, each scored 1 to 5 with a short quote from the call. The interview counts double next to the CV and profile scores in the overall result.
Admin report and candidate history
Summary, call analysis, CV analysis and profile analysis tabs, the recording and full transcript, and every earlier attempt by the same phone number with its date and score.
WhatsApp follow-through
Candidates receive a short result page to share, interview times are confirmed on WhatsApp, and a cancelled interview sends a ready message automatically.
Technologies & capabilities
- Next.js
- Supabase / PostgreSQL
- Realtime AI voice
- AI scoring with structured output
- WhatsApp Business API
- English / Arabic
- Mobile-first PWA
How we delivered
Built With the Hiring Manager, Changed by Real Calls.
- 01
Design session
We mapped the application steps, the call scenario and the ten questions with the hiring manager.
- 02
First version in days
A working wizard, AI call and report were live quickly, so real candidates could use it.
- 03
Feedback rounds
Each round of manager feedback changed the form, the AI client's behaviour, the timer and the report.
- 04
Fixes from the field
Because some networks restrict direct browser voice connections, we routed the audio through our own server and re-tested.
Result
The First Interview Now Happens Without the Team.
Every applicant is tested the same way
Same questions, same scoring, a different client each time, so results can be compared fairly.
Managers read reports, not CV piles
The panel shows who is ready to hire, who to invite to an interview and who to invite to training.
Progress over time
Internal sales staff can retake the test, so the team sees how scores change before and after training.
Less manual messaging
Interview confirmations and cancellations go out on WhatsApp from the panel.
Related services
- Custom Software DevelopmentBusiness software, CRM systems, dashboards and portals built around the way your company actually works.
- AI & Business AutomationAutomate approvals, back-office tasks, documents and data transfer between your systems, with AI where it helps.
- AI Agents & WhatsApp AutomationAn Arabic and English AI assistant on WhatsApp that answers, qualifies leads, updates your CRM and follows up.
FAQ
Frequently asked questions
Let’s Talk
Want to Screen Candidates Before They Reach Your Team?
Tell us how you hire today and we will show you what an AI-assisted process could look like.
