The first person who reads your job application in 2026 is probably not a recruiter. It's an algorithm. For international IT graduates chasing a 485-to-186/190 pathway, that algorithm is the first gate — and getting bounced there means the human never sees your file at all.
Two platforms show up repeatedly in Australian corporate and graduate hiring: Sapia and HireVue. They look similar from the outside — complete a short interview, wait for a score — but they work very differently. That difference matters if English isn't your first language, because one of them literally cannot hear your accent.
Sapia: Text only. Your accent is not in the model.
Sapia runs what it calls a Chat Interview — a text conversation where you type answers to 8–12 behavioural questions. No video. No audio. You never press record.
The scoring happens through SAIGE™, an LLM-based engine trained on over 8 million real structured interviews and 3 billion words of candidate text across 77 countries (sapia.ai/platform). It doesn't hear your accent, evaluate your eye contact, or penalise you for a grammatical slip you'd only make when speaking. It reads your written answers and evaluates them against the behavioural competencies the employer defined before the interview opened.
That architecture is important. Qantas deployed Sapia for their 2026 graduate cohort — 4,500 applicants completed the chat interview, and 96.7% of them finished it, rating the experience 8.8 out of 10 (sapia.ai/blog/graduate-hiring-qantas). Michael Eizenberg, Head of Talent at Qantas, described it as "a valuable tool for improving diversity, accessibility and efficiency." The hiring team reported equitable outcomes across gender and ethnicity.
The external bias audit backs this up. BLDS — an independent assessment firm — ran Standardised Mean Difference (SMD) tests across 23 assessments and Adverse Impact Ratio (AIR) tests across 49 scenarios. Finding: "no evidence of practically significant disparate impact for any protected group" (sapia.ai/sapia-ai-recruitment-software/sapia-releases-its-independent-bias-audit). The audit covers US and Canada cohorts, not Australia specifically, and doesn't break out language background as a protected category — so treat it as partial evidence, not a guarantee. But the point stands: there is no audio or video channel through which your accent can enter Sapia's scoring.
What Sapia actually scores: Job-relevant behavioural competencies. For the Qantas grad program, those were curiosity, adaptability, and empathy. SAIGE looks at how you frame situations, whether your story shows initiative, how clearly you separate what you did from what the team did, and whether there's a concrete outcome at the end. STAR structure — Situation, Task, Action, Result — is the format the system is trained to reward.
HireVue: Video, but not what you might fear
HireVue is more complex. It offers five distinct assessment types (hirevue.com/platform/assessments):
- AI-Scored Interviews — video responses, under 20 minutes
- Game-Based Assessments — non-verbal psychometric tasks
- Virtual Job Tryout — scenario simulations
- Technical Assessments — auto-graded coding evaluations
- Language Proficiency Tests — a separate, opt-in product
That last item is the one to notice. Language Proficiency Tests are a distinct product that employers deliberately purchase and activate. They are not present in every HireVue deployment. If you're doing a standard AI-Scored Interview, you are not in a language proficiency test.
Game-Based Assessments are explicitly available in multiple languages (hirevue.com/platform/game-based-assessments), and measure three trait clusters: Personality & Work Style, Working with People, and Working with Information. These are cognitive and behavioural signals — not English proficiency.
The AI-Scored Interview is where the video happens. HireVue describes its goal as assessing "how well you can communicate and process your thoughts" — behavioural competencies similar to what a structured human interview would probe. HireVue does not publish a full technical breakdown of its video scoring weights, so I won't fabricate one. What's clear from their published documentation is that the platform measures competencies tied to specific job requirements, not generic personality. The content of what you say — structured, specific, evidence-based — is the primary signal.
The practical implication: if you're sitting a HireVue video interview and your main worry is your accent, you're focused on the wrong variable. A rambling, vague answer in native-sounding English will score lower than a direct, structured answer in clear but accented English. The engine is evaluating your response for behavioural content, not phonetics.
How to prepare for each tool
For Sapia
- Write full sentences, not chat messages. Sapia is text-based, but it's not a casual conversation. Draft your response in a notes app first, then paste it in.
- Use STAR explicitly. "In my final-year capstone at UNSW, the client shifted scope two weeks before delivery. I owned the API integration layer, so I proposed dropping the reporting module and got their written agreement in 24 hours. We shipped on time and they extended the engagement the following semester."
- Use the untimed format. Sapia interviews have no countdown timer on individual questions. Spend 5 minutes drafting each answer before submitting.
- Target 150–250 words per answer. Enough specificity to show a real situation. Not so long that you bury the signal in padding.
- Spell-check before submitting. The model reads text. Systematic typos don't help your professional signal.
For HireVue
- Identify which product type you're sitting. The invitation email or brief usually names it. AI-Scored Interview, Game-Based, and Technical each require different preparation.
- Practise recording yourself before the real thing. Not to perfect delivery — to notice if you're rambling. Each video answer should run under two minutes.
- Apply the same STAR structure as you would for Sapia, now spoken aloud. Keep outcomes concrete: "deployed to production," "merged after two rounds of review," "cut build time by 30%."
- Check whether Language Proficiency is in scope. If the brief mentions it as a distinct component, prepare specifically for it — it will test your English communication ability directly.
- Don't overthink Game-Based. These measure stable cognitive traits. Familiarise yourself with the format (attention, pattern recognition, working memory tasks) but don't try to game them — the traits they target don't shift with short-term prep.
What passing AI screening unlocks on the 485 pathway
AI screening is the gate before the human conversation. Qantas had 4,500 graduate applicants for a cohort that probably numbered in the dozens. The algorithm's job is to get that number to something a recruiter can actually review. If your Sapia responses read as low-competency-signal, your resume never opens.
The good news is that neither Sapia nor HireVue is penalising your English accent by design. The mechanism isn't there in Sapia (text only). In HireVue the primary scoring signal is behavioural content. The real risk for international graduates is generic answers — "I worked collaboratively and communicated effectively with stakeholders" gives the system nothing to evaluate. "I caught a schema mismatch between our dev and prod databases 48 hours before go-live, flagged it to the tech lead, rewrote two migration scripts, and documented the fix in the post-incident review" gives it something to score.
Your 485 visa gives you the time to build that story base. Every project you complete, every production incident you help contain, every scope negotiation you participate in becomes material for these answers. The candidate who can answer "tell me about a time you had to push back on a deadline" with a specific, quantified outcome clears AI screening regardless of accent.
Once you're through AI screening and in front of a human interviewer, the visa conversation can happen. That's where you explain your 485 status and your pathway to 186 or 190. The AI screen just has to not stop you getting there.
Use Gradland's Interview Prep tool to practise STAR-format answers against real behavioural questions — the same structure Sapia and HireVue reward.
For current occupation list data and 485 nomination thresholds by state, the AU Insights section is updated as Home Affairs releases changes. The Visa News feed covers Home Affairs announcements relevant to 485, 186, and 190 applicants as they happen.