Frequently Asked Questions

Traditional AI tests use static multiple-choice questions that candidates can Google in seconds. AISA runs a live, adaptive conversation that follows up on what the candidate actually says — probing deeper on strong claims and pivoting when answers are vague. This makes it far harder to fake and far more reflective of real-world AI ability. Instead of testing recall, AISA tests whether someone can actually think with AI. Read more in our guide: Beyond Multiple Choice (/resources/beyond-multiple-choice).

AISA evaluates practical AI fluency across 5 dimensions and 11 criteria. The dimensions are Prompting & Communication, Critical Thinking, Technical Understanding, Workflow & Application, and Safety & Responsibility. Each dimension captures a different facet of how someone works with AI — from writing effective prompts to recognising model limitations to integrating AI into daily workflows. The result is a comprehensive profile, not just a single number. See the full breakdown at /resources/the-aisa-rubric.

Instead of a quiz with predetermined answers, candidates chat naturally with Aisa, our AI facilitator. Aisa adapts in real time — asking follow-up questions, probing claims with scenarios, and moving on when a topic is well covered. This mirrors how people actually demonstrate expertise: through conversation, not checkboxes. The format also lets candidates showcase strengths that rigid tests would miss entirely.

Most assessments take between 20 and 45 minutes, though the conversation can run up to 60 minutes if the candidate wants to go deeper. There is no strict time limit — Aisa adapts pacing based on how much evidence has been collected across all 11 criteria. A short, high-quality session can produce a perfectly valid report.

AISA adapts its conversation to any knowledge-worker role. It adjusts the topics, scenarios, and depth of probing to match the role context — a developer gets different questions than a product manager or designer. We have tailored assessment paths for developers, product managers, designers, data scientists, and a general track for all other roles. Learn more at /roles.

Every score is tied directly to evidence from the conversation — specific quotes and demonstrated behaviours, not subjective impressions. A second AI model independently evaluates the entire transcript, and a calibration pass adjusts for scoring consistency. This dual-track architecture means two candidates who say the same things will receive the same scores, regardless of when or how they take the assessment.

The report includes a composite score out of 100, individual scores for each of the 11 criteria, a persona classification (like The Architect or The Builder) that captures the candidate's AI working style, a personalised deep dive with specific evidence from the conversation, an action plan with growth recommendations, and a learning box with next steps. It is designed to be useful for both hiring decisions and personal development.

AISA classifies each candidate into one of 10 AI personas based on their scoring profile — ranging from The Bystander (limited AI engagement) to The Oracle (deep, strategic AI fluency). Each persona captures a distinct working style and relationship with AI tools. Personas like The Builder, The Architect, and The Strategist highlight different strengths. Employers find them useful for understanding team composition, and candidates find them motivating. Explore all 10 at /resources/ai-persona-types.

One detailed report is generated and shared equally with the candidate and the employer. There are no separate views or hidden sections — everyone sees the same evidence-based scores, the same persona, and the same recommendations. This transparency builds trust and ensures fairness on both sides.

AISA has built-in protections against gaming. It detects copy-paste behaviour, sudden style shifts, AI-generated responses, and suspicious response speed through a multi-metric AI fingerprint system. Because the conversation is adaptive and unpredictable, there is no answer key to memorise. Flagged anomalies appear directly in the report with full transparency. Read more about our approach: /resources/beyond-multiple-choice.

AISA tracks five behavioural metrics in real time: correction rate, typing density, message length patterns, formality shifts, and response uniformity. Together these form an AI fingerprint that distinguishes natural human responses from AI-generated ones. The system also detects dictation (voice-to-text) and adjusts accordingly, so it does not penalise candidates using accessibility tools.

Yes. AISA evaluates AI skills and reasoning, not English fluency. The scoring rubric explicitly separates communication clarity from language proficiency. Aisa adapts its language complexity based on the conversation, and scoring is calibrated to reward substance over style. We continuously monitor for language bias in our scoring pipeline.

Sign up as an employer, purchase credits, and send assessment invitations to candidates via email. Each candidate completes a conversation with Aisa, and you receive a detailed report with scores, evidence, persona classification, and recommendations. Many employers use AISA as a screening step before technical interviews, saving hours of manual evaluation. Learn more at /employer.

Absolutely. Many organisations use AISA to benchmark their current workforce — identifying skill gaps, planning targeted training, and tracking improvement over time. The assessment works the same way: each team member takes a conversational assessment and receives a personalised report with growth recommendations.

Asking a few AI questions in an interview gives you a rough impression but no standardised data. AISA evaluates 11 specific criteria with consistent rubric-based scoring, produces evidence-backed reports, and enables apples-to-apples comparison across every candidate. It also covers dimensions interviewers rarely probe — like safety awareness, critical evaluation of AI outputs, and workflow integration patterns.

Each assessment uses one credit. Credits start at $10 each, with volume discounts for larger packs. There is no subscription, no monthly minimum, and no cancellation process. You buy credits when you need them and use them at your own pace. Unused credits never expire.

Yes — and we encourage it. Sign up as an individual professional to take the assessment yourself and see exactly what your candidates will experience. You will receive a full report with your own AI persona, scores, and personalised recommendations. You can also view a public sample report from the landing page to see the output format.

In under two minutes. Sign up, purchase a credit pack, and send your first assessment invitation immediately. There is no setup wizard, no integrations required, and no IT involvement needed. Candidates receive a magic link via email and can start their assessment right away.

We do not offer a free trial, but you can view a detailed sample report on our landing page to understand exactly what the assessment produces. The $10 single-credit option is designed to be a low-risk way to experience the full assessment firsthand.

Yes. All conversation data is encrypted in transit (TLS) and at rest. Data is used solely for generating the assessment report and is never sold, shared with third parties, or used to train AI models. We follow strict data minimisation principles — we only collect what is needed for the assessment. You can request full data deletion at any time.

Assessment data is stored in a Supabase-hosted PostgreSQL database with row-level security policies, hosted in the EU (AWS eu-west-1). The application is deployed on Vercel's edge network. All infrastructure providers are SOC 2 compliant. For full details, see our privacy policy at /privacy.

None at all. AISA uses a simple pay-as-you-go credit model. You buy credits when you need them, use them whenever you want, and stop whenever you want. No subscription, no annual commitment, no cancellation process. Unused credits never expire.