ai interviews.

Role

Lead UX Designer

Company

Naukri.com / Doselect

Timeline

2024 · Ongoing

Tools

Figma · Jira

From recruiter interviews and sales feedback, we observed that early interview rounds were increasingly being deprioritized by hiring managers.

• 7 out of 10 recruiters said first-round interviews mostly focused on basic validation like communication, fundamentals, and intent
• Most recruiters estimated that despite being simple, these rounds consumed roughly one-third of total interviewer time per role
• Several recruiters shared that senior hiring managers frequently delayed or skipped first rounds, directly impacting time-to-hire

In parallel, recruiters mentioned that DoSelect's Manual Interviews were increasingly replaced by Google Meet / MS Teams with plugins, which led to:

• Lower repeat usage of Manual Interviews
• Reduced perceived differentiation from generic meeting tools
• Drop-offs after interview scheduling

Together, recruiters described first-round screening as a high-effort, low-value bottleneck, making it a strong candidate for automation.

Manual Interviews reinforced these problems by requiring constant hiring manager involvement, making the system slow, dependency-heavy, and difficult to scale.

Based on recruiter expectations shared during interviews, we defined success as:

• Reducing recruiter coordination effort in early screening
• Bringing first-round turnaround time down from days to hours
• Standardizing evaluations so candidates could be compared reliably
• Improving candidate completion through asynchronous, flexible interviews
• Retaining recruiters within the DoSelect ecosystem instead of external tools

Before designing solutions, we aligned on a few key research questions:

• Why did recruiters stop using Manual Interviews after initial adoption?
• Which interview steps required the most effort but contributed the least to decision confidence?
• Where were recruiters comfortable allowing AI to replace human effort?

A recurring theme across interviews was reducing dependency on hiring managers for question creation (which often delayed the process by 2–3 days) and first-round evaluation, which was often rushed or skipped.

recruiter interviews.

10+ recruiters from active DoSelect customers · 6 recruiters from churned or low-usage accounts

stakeholder discussions.

Product and Engineering teams · Sales feedback from lost or stalled deals

internal data reviews.

Interview setup drop-off points · Average time-to-hire by role type

introductory screening is the highest ROI use case for AI.

• ▸Around 7 out of 10 recruiters said they were comfortable using AI for first-round screening
• ▸Only 1–2 recruiters expressed comfort with AI making final hiring decisions

hiring managers actively avoid early-stage efforts.

• ▸More than 8 out of 10 recruiters said managers did not want to write interview questions
• ▸Roughly two-thirds said managers preferred not to evaluate early-stage candidates

bulk hiring amplifies inefficiencies.

• ▸Recruiters handling 50–200 candidates per role were the most vocal about delays and fatigue
• ▸These recruiters perceived AI-led screening as the highest value

trust is non-negotiable.

• ▸Nearly every recruiter interviewed raised concerns around impersonation, plagiarism, or misuse
• ▸Proctoring was consistently described as a baseline requirement

Based on recruiter confidence levels, we committed to an MVP-first, scalable approach focused on automating introductory screening, covering roughly 60% of early-stage effort.

Key decisions informed by recruiter interviews:

• AI Interviews handled introductory validation only
• Questions were dynamically generated using role, experience, and skill expectations
• Evaluations were structured and explainable, with scorecards preferred over free-text
• Recruiters could complete setup independently, in under 10 minutes

We chose a conversational chatbot flow over forms because recruiters naturally described hiring needs conversationally and found long forms exhausting during high-volume hiring.

What interviews and testing revealed:

• Recruiters thought through role expectations before exact questions
• 2 out of 3 recruiters reported form fatigue
• Progressive refinement of inputs was preferred
• A DoIQ-powered chatbot enabled ~30% faster setup, fewer post-creation edits, and higher confidence in AI-generated questions

To assist recruiters in faster, more objective decisions, we designed a smart evaluation layer post-interview completion.

Early candidate feedback influenced refinements to the AI agent:

• Several candidates expressed nervousness when interacting with AI
• Pauses were often perceived as failure
• Overly robotic responses reduced comfort

Key UX decisions included:

• Regional tone and friendly phrasing
• Smart fillers during pauses
• No answer leakage
• Optional feedback on request

Recruiters later reported fewer candidate complaints post-interview.

what worked well.

• ▸End-to-end interview creation without hiring manager involvement
• ▸Structured AI-driven scoring improved decision clarity
• ▸Chat-based creation significantly lowered setup friction
• ▸Proctoring increased trust in AI-led interviews

open challenges & learnings.

• ▸Some candidates remained skeptical of AI evaluation
• ▸A few recruiters raised concerns around transparency

future opportunities.

• ▸Stronger explainability of AI scoring
• ▸Hybrid AI + human validation for borderline candidates