Why pre-screen HR chatbot designers before the portfolio interview
An HR assistant is a harder product than a support bot because the questions it receives are confidential, sometimes distressing, and frequently outside what any model should answer. A designer who has only built demos will show you a smooth flow for booking annual leave. What you need to know is whether they have thought about the employee typing a grievance at midnight, and whether the bot they shipped was still in use six months later. Neither shows up in a portfolio.
What actually matters when screening AI-Powered HR Chatbot Designer candidates
- 01
Portfolio
Ask for shipped conversational flows: onboarding FAQ bots, leave-balance lookups, benefits enrolment journeys. Look for containment rates, deflected ticket volumes, and the platform used (Dialogflow CX, Rasa, Copilot Studio).
- 02
Craft and rationale
Probe persona and tone decisions, prompt or utterance design, fallback and disambiguation logic, and how they handle sensitive HR topics like grievances or pay disputes.
- 03
Feedback and iteration
Test how they used transcript review, unmatched-utterance logs, and A/B tests of prompt wording to rework flows after launch, plus response to HRBP and works council pushback.
- 04
Working with the brief
Judge intake with HR stakeholders: mapping policy documents and HRIS data (Workday, SuccessFactors) into intents, scoping what the bot must never answer, and agreeing success metrics.
Pre-screening questions to ask AI-Powered HR Chatbot Designer candidates
12 questions grouped by what they test. Ask the same set in every screen and score answers on a consistent scale, or send them as an async video screen and compare answers side by side.
Bots that shipped
3 questions01Describe a time you implemented an AI-powered chatbot for an HR department.
Listen forA named deployment with workforce size and the intents it actually covered, plus whether it was still running when they left the project.
Describes a prototype or pilot that never reached employees, or cannot say what the bot was used for most.
02What experience do you have integrating chatbots with HR systems such as Workday or ADP?
Listen forA real integration with authentication and permissions handled, plus what the bot was allowed to change rather than only read.
Integration limited to static content, or no view on how the bot verified who it was speaking to.
03Which AI frameworks and technologies are you most proficient with?
Listen forTools tied to a shipped product with a reason for the choice, including where they used retrieval rather than letting a model answer freely.
Lists current model names with no deployment behind them, or no distinction between generated and retrieved answers.
Conversation design reasoning
3 questions04How do you approach user experience design for conversational interfaces?
Listen forDesign decisions for the unhappy path: how the bot handles being misunderstood, repeats itself differently, and admits it cannot help.
Describes only the intended flow, with no plan for misunderstanding or for a user who goes off script.
05How do you maintain a consistent tone of voice in chatbot responses?
Listen forA written voice specification with examples, plus how they keep generated responses inside it rather than relying on the model to behave.
Tone handled entirely by a system prompt with no review, or no distinction between routine and sensitive messages.
06Describe a scenario where your chatbot design had to support multiple languages.
Listen forReal multilingual work with the hard parts named: idiom, formality levels, and how they validated quality in a language they do not speak.
Machine translation applied to the whole flow with no native review, or no awareness of formality registers.
Improving after launch
3 questions07How do you measure the effectiveness of an HR chatbot once it is deployed?
Listen forContainment or resolution rate with a figure, alongside a measure of whether employees got the right answer rather than just an answer.
Measures conversation volume alone, or has never checked whether the answers given were correct.
08How do you ensure a chatbot keeps improving after its initial deployment?
Listen forA review loop over failed conversations, with intents added and removed based on what employees actually asked rather than what was planned.
Treats launch as the end of the work, or improvement described entirely as retraining on more data.
09What is your process for conducting user testing with HR chatbots?
Listen forTesting with real employees rather than colleagues, including deliberately awkward phrasings, and a change they made because a session went badly.
Testing done only within the project team, or scripted sessions where testers were told what to type.
Escalating to humans
3 questions10What methods do you use to ensure a chatbot handles sensitive and confidential information securely?
Listen forData minimisation, retention limits and identity verification, plus a clear rule about what the bot must never store or repeat back.
Relies on the platform's security claims, or logs full conversations indefinitely with no retention policy.
11Have you faced compliance or legal issues on HR chatbot projects? How did you address them?
Listen forA specific constraint they designed around, such as works council consultation or data residency, and what it changed about the build.
Compliance treated as a review someone else performs, with no example of a design changed because of it.
12What strategies do you use so a chatbot can handle a wide variety of employee queries?
Listen forA deliberate scope with an explicit handover for anything outside it, including distressing disclosures routed to a person immediately.
Aims for the bot to answer everything, or has no defined path for a grievance or wellbeing disclosure.
How to score responses
Score every candidate on the same four criteria immediately after the screen. At this stage you are shortlisting for panel interviews, not making the final call.
Portfolio
35%5Walks through two or three live HR bots with named intents, containment or CSAT figures, and honest notes on flows that failed.
Craft and rationale
25%5Explains persona choices, escalation thresholds, and error-recovery copy with reasons tied to employee trust rather than personal taste.
Feedback and iteration
25%5Cites specific transcript findings that changed a flow, and describes rewriting copy after legal or HR review without defensiveness.
Working with the brief
15%5Turns vague requests such as "reduce HR emails" into a scoped intent list, escalation rules, and measurable containment targets.
A demo flow for booking leave looks identical to a bot that survived contact with a real workforce. A one-way video screen lets you hear which one a candidate has built.
Try it on HirevireScreening FAQ
Process basics
How long should a pre-screening round for an HR chatbot designer take?
Fifteen minutes across eight to ten questions, answered async. Enough to confirm they have shipped to real employees, hear their approach to confidentiality and escalation, and check HRIS integration experience before a portfolio interview.
Should I weight AI framework knowledge or conversation design?
Conversation design, for this role. Model and framework choices move quickly and are learnable. Knowing which queries a bot should refuse, and how to hand over without making the employee repeat themselves, is the part that decides whether people keep using it.
Evaluating answers
What is the strongest signal when screening an HR chatbot designer?
How they handle a sensitive disclosure. Designers who have shipped into a workforce have already had to decide what the bot does when someone reports harassment or asks about a mental health issue. Candidates who have not will treat it as an edge case.
How do I check that a bot they built was actually used?
Ask for containment and usage numbers rather than launch dates. A designer who stayed with the product can tell you what proportion of queries were resolved without a human, how it moved, and which intents they removed because nobody asked them.
Go deeper on this role
Sanat has been hiring since 2012 and watching the recruitment industry change up close ever since, and turned that screening process into Hirevire's video screening platform. LinkedIn
