TalentReef – AI-Assisted Applicant Review
Using attitudinal and behavioural research to determine where AI could reduce friction in high-volume applicant review.
Senior Product Designer
Mitratech
AI-Assisted Decision Workflows
Recruiters & Hiring Managers
The Challenge
Before adding AI, I needed to understand what was actually broken.
TalentReef was exploring its first AI-powered capabilities within the applicant review experience.
The existing Applicant View contained valuable information across multiple tabs, including Overview, History and Applications.
Hiring Managers were repeatedly moving between these areas to build a complete picture of an applicant.
The question wasn't simply:
"Where should we put AI?"
It was:
"What is missing from the review experience that AI could meaningfully improve?"
Why It Mattered
Hiring Managers work through large applicant volumes and need to make decisions quickly.
If AI simply added another layer of information to an already fragmented experience, it would increase complexity rather than reduce it.
The opportunity was therefore to use AI to consolidate decision-relevant information at the moment of review.
Success Criteria
The design needed to:
Identify the highest-value gap in applicant review.
Validate the gap with real users.
Confirm whether observed behaviour matched reported friction.
Determine the most appropriate location for AI-assisted information.
Make AI insights understandable and connected to underlying applicant data.
Preserve human judgement rather than replacing it.
Research & Discovery
Two research methods. One behavioural story.
01 — Pendo Poll
I designed and launched a targeted Pendo Poll to understand what Hiring Managers and HR Admins considered the biggest gaps in the existing applicant review experience.
Research data:
11/34 — 32%
identified prior/duplicate applications as the biggest gap.
7/34 — 21%
struggled to understand applicants at first glance.
7/34 — 21%
struggled to prioritise applicants.
7/34 — 21%
struggled to review details without switching sections.
Then:
02 — Session Replays
I analysed Pendo Session Replays to validate whether reported navigation friction was reflected in actual behaviour.
The replays confirmed repeated tab-switching during applicant review.
Key Insight
"Users weren't asking for AI. They were asking for consolidation."
AI became useful only after the underlying information architecture problem was understood.
Design Thinking
Research before AI assumptions.
Consolidate before adding intelligence.
Put AI where the user already begins the task.
Make AI insights contextual and explainable.
Keep humans in control of decisions.
Validate what users say against what they actually do.
Design Decisions
01 — AI Insight in Overview
The Overview tab became the proposed home for AI-generated applicant insights because it is the first surface visible when the drawer opens.
02 — Open by default
The panel opens without requiring another interaction, reducing friction at the highest-value moment.
03 — Consolidate key signals
The AI layer brings important applicant signals together instead of asking managers to navigate between tabs.
04 — Explain the recommendation
AI insights should not simply state a conclusion. They should connect the conclusion to the underlying applicant information.
For example, instead of:
"Candidate is a good fit."
the experience should explain why the candidate received that assessment and expose the supporting signals.
AI Trust/System Design
AI should explain, not obscure.
Contextual
Connect insights to applicant data.
Explainable
Show the signals contributing to an assessment.
Actionable
Help the recruiter decide what to do next.
Note: Employment history was not collected for this application workflow.
Final Solution
A research-grounded AI Insight panel positioned inside the Overview experience, designed to consolidate the most important applicant signals without creating another destination for managers to navigate.
Use cases/edge cases defined
Outcome/Validation
Research validated placement
Pendo Poll + Session Replay synthesis presented and approved
Behavioural pattern confirmed
tab-switching during applicant review validated across multiple session replays
AI integration direction established
Overview tab, open by default, key info consolidated
What I'd Do Next
Usability test the AI Insight panel with real Hiring Managers.
Validate whether default-open feels useful or intrusive.
Track tab-switching behaviour post-launch.
Measure AI-panel engagement.
Validate whether managers understand the rationale behind AI-generated insights.
Reflection
The most important AI design decision wasn't where to put the AI. It was deciding not to design it until the underlying user problem was understood.
View All Work