C E R T I F I C A T I O N G U I D E Qlik • QAIS • AI Specialist Qlik AI Specialist Certification Guide: QAIS Syllabus and Questions Syllabus, sample questions and a study plan for the Qlik AI Specialist (QAIS) exam Inside: the exam fact sheet checked line by line against Qlik’s own exam-details page, all five exam topics expanded to every published objective, why three fifths of this paper is not about Qlik at all, two things that have quietly changed since the exam launched, what the free sample set covers and the two topics it never touches, a five-step route to exam day, and ten sample questions with a full answer key. T H E B L U E P R I N T Introduction to AI 30% Business applications 30% Qlik Answers 15% Qlik Machine Learning 15% Insight Advisor 10% E X A M A T A G L A N C E 90 MINUTES 50 QUESTIONS 73% TO PASS 5 EXAM TOPICS 17 OBJECTIVES Prepared by AnalyticsExam • www.analyticsexam.com www.analyticsexam.com Qlik • QAIS Qlik AI Specialist Certification (QAIS) 1 CERTIFICATION GUIDE Contents Ctrl+click any line to jump straight to that section. The same seven entries appear in your PDF reader’s bookmark pane. SECTION 01 Exam Overview 2 SECTION 02 Two Things That Have Quietly Changed 3 SECTION 03 The Five Exam Topics, Objective by Objective 4 SECTION 04 What the Credential Is Worth 6 SECTION 05 Getting Certified, Step by Step 7 SECTION 06 QAIS Sample Questions 7 SECTION 07 Where to Go Next 11 www.analyticsexam.com Qlik • QAIS Qlik AI Specialist Certification (QAIS) 2 SECTION 01 Exam Overview The Qlik AI Specialist (QAIS) certification exam is the newest credential Qlik offers and the only one that is mostly not about Qlik. Qlik’s own description of it is short: the certification "validates your knowledge of key AI concepts and applications, including the fundamental skills to create Qlik predictive AI models and generative AI assistants". Read the blueprint and the emphasis lands in an unexpected place — three fifths of the paper is general AI literacy that would sit just as well on a vendor-neutral exam, and only two fifths touches a Qlik product at all. Every figure below comes from the official Qlik QAIS exam-details page or, where Qlik is silent, from the AnalyticsExam QAIS syllabus page, and each row says which. Certification Qlik AI Specialist Certification Exam code QAIS Duration 90 minutes, per Qlik’s current exam-details page — two other sources still say 120; see Section 02 Number of questions 50, per Qlik Passing score 73%, per Qlik Exam fee USD 250, per the AnalyticsExam syllabus page — Qlik’s exam page states no price and links out to its own purchase store Exam topics Five, each with a published percentage weighting Published objectives 17 across the five topics Booking Through Qlik Learning — but read the note on Qlik’s Pearson VUE link in Section 02 before you follow it Credential awarded A Qlik AI Specialist Certification digital badge issued through Credly; the badge page states no expiry date Stability caveat Qlik states that "Exam content is updated periodically. The number and difficulty of questions may change. The passing score is adjusted to maintain a consistent standard." Page published Qlik dates the exam-details page 13 May 2025 The attributions in that table are deliberate, and so are the rows that are missing from it. Qlik’s exam-details page publishes three numbers — duration, question count, pass mark — plus the five topics and their weightings, and then it stops. It names no prerequisite, no language list, no validity period and no delivery or proctoring arrangement. A guide that filled those rows in with "None" or with a plausible guess would be inventing them, so they are simply absent here. The only figure in the table that is not Qlik’s is the fee, and it is labelled as such. www.analyticsexam.com Qlik • QAIS Qlik AI Specialist Certification (QAIS) 3 Taking the current numbers at face value, 50 questions in 90 minutes is 108 seconds each , and a 73% pass mark means 37 of 50 correct . That pass mark is the number to sit with. It is materially higher than most analytics certifications — Qlik’s own Sense Business Analyst exam passes at 62% — and it leaves you thirteen wrong answers across the whole paper. On a blueprint this broad, thirteen is not a lot of room. The arithmetic worth knowing before you book 73% of 50 is 37 correct. You can afford 13 wrong answers in total. Topics 1 and 2 are 60% of the paper: about 30 questions, none of them Qlik-specific. The three Qlik product topics together are 40%: about 20 questions. Insight Advisor, the smallest topic, is 10% — roughly five questions. That split is the single most useful fact about QAIS, and it cuts both ways. If you already work in Qlik every day, the majority of this exam is material you will not pick up from the product, and the Qlik AI Certification Exam Preparation path is where Qlik points candidates for it. If you arrive from an AI or data science background instead, the general half will feel straightforward and the 40% of product questions is the part that needs hands-on time in a Qlik Cloud tenant. SECTION 02 Two Things That Have Quietly Changed QAIS is a young exam — Qlik announced it on the Qlik Community in June 2025 and dates the exam-details page 13 May 2025 — and two things about it have moved since launch without much announcement. Neither changes the blueprint, but both will trip up a candidate preparing from older material. First, the duration. Qlik’s current exam-details page says 90 minutes . Qlik’s own launch announcement, posted in June 2025, says 120 minutes, and the AnalyticsExam syllabus page still says 120 as well. Same 50 questions and same 73% pass mark in all three. The most likely reading is that the exam was shortened after launch and the announcement post was never revised, which would make the live exam-details page the current figure — that is why this guide leads with 90. But this guide is not in a position to prove that, and the difference is half a minute per question, which is the difference between comfortable and rushed. Check the duration on Qlik’s exam page on the day you book , and if you practise, practise at the 90-minute pace rather than the generous one. Second, the product in topic 4 has been renamed. The blueprint says "Qlik AutoML" in all three of its objectives. Qlik’s current documentation for that product is headed machine learning with Qlik Predict, and introduces it as "Automate machine learning for your analytics team with Qlik Predict. With the simple code-free interface, you can easily create machine learning experiments to generate models and make predictions." The help URLs still contain automl , and partner and community material routinely writes it as "Qlik Predict (formerly Qlik AutoML)". Treat the two names as the same product. If a question in the exam says AutoML and everything you revised said Predict, nothing has gone wrong. One more caution that is not a change so much as a dead end. The exam-details page carries a "Register Here" link pointing at Pearson VUE’s Qlik page. That page now answers with "Pearson VUE no longer delivers exams for the testing program you are trying to reach" and refers candidates back to the www.analyticsexam.com Qlik • QAIS Qlik AI Specialist Certification (QAIS) 4 program. Do not plan your booking around Pearson VUE. Buy and schedule through Qlik Learning’s own purchase and registration flow, reached from the exam-details page, and expect the delivery arrangements to be described there rather than by a third-party test-centre network. Because Qlik does not state a delivery method or a proctoring arrangement anywhere on the exam page, this guide does not claim one. SECTION 03 The Five Exam Topics, Objective by Objective Qlik publishes an exact percentage for each topic rather than a range, so the chart below is the real blueprint and not a proxy for it. The objective lists beneath are reproduced as Qlik words them. Introduction to AI 30% Business applications 30% Qlik Answers 15% Qlik Machine Learning 15% Insight Advisor 10% Published weightings from the Qlik AI Specialist Certification Exam Details page. Two topics tie at the top, which is unusual and worth pausing on. Introduction to Artificial Intelligence and Business applications for Artificial Intelligence are 30% each, and between them they carry 60% of the paper. Neither mentions a Qlik product. This is an AI literacy exam with a Qlik module bolted on, not a Qlik exam with an AI flavour, and the practical consequence is that a Qlik consultant of ten years’ standing can walk in and fail it. The three product topics — Qlik Answers, Qlik Machine Learning and Insight Advisor — total 40%, which on its own is well short of the 73% you need. 1 Introduction to Artificial Intelligence — 30% of the exam, 3 objectives • Understand various subsets of AI including GenAI and machine learning • Contrast the different ways humans communicate with AI (Natural Language Processing, Large Language Models) • Define common AI terms and concepts 2 Business applications for Artificial Intelligence — 30% of the exam, 6 objectives • Identify appropriate use cases for GenAI and machine learning • Recognize the LLM Application Project Lifecycle • Evaluate the production of LLM apps including adapting data pipelines, reducing latency, and expanding the usefulness of LLMs • Explain the fundamentals of the Machine Learning workflow • Assess Data Governance, security practices, and ethical considerations for projects involving AI • Identify the limitations of AI and the data challenges involved in implementing GenAI www.analyticsexam.com Qlik • QAIS Qlik AI Specialist Certification (QAIS) 5 3 Fundamentals of Qlik Answers — 15% of the exam, 3 objectives • Outline a typical workflow for Qlik Answers • Recognize key concepts and terms associated with Qlik Answers, including retrieval augmented generation (RAG) • Evaluate use cases for Qlik Answers 4 Fundamentals of Qlik Machine Learning — 15% of the exam, 3 objectives • Understand the foundations of Qlik AutoML • Demonstrate appropriate application of the Qlik AutoML workflow • Evaluate use cases for Qlik AutoML 5 Fundamentals of Insight Advisor — 10% of the exam, 2 objectives • Generate insights using Insight Advisor • Demonstrate best practices in interacting with Insight Advisor (prompt generation) Read the verbs and the weightings together and the study plan writes itself. Topic 1 asks you to understand , contrast and define — that is vocabulary and hierarchy. You need the nesting of AI, machine learning, deep learning and generative AI straight, the difference between natural language processing as a field and large language models as a technique, and the everyday terms: training versus inference, corpus, token, parameter, hallucination. It is 30% of the paper for what is essentially a glossary, and sample questions 4, 6 and 7 in this guide are exactly that shape. Topic 2 shifts to identify , recognize , evaluate and assess , and it is where the exam gets its judgement questions. The LLM Application Project Lifecycle is named explicitly as something to recognise, so learn it as an ordered set of stages — scope the use case, select and adapt a model, evaluate, deploy and monitor — because question 8 asks you to place an activity in it. The production objective is similarly concrete: Qlik lists "adapting data pipelines, reducing latency, and expanding the usefulness of LLMs" as three distinct levers, and question 2 turns on picking the right one for a stated symptom. The governance and limitations objectives are the two that reward reading widely rather than product documentation. The three product topics each follow the same three-part pattern — outline the workflow, know the key terms, evaluate the use cases — and the third part is where the marks actually move. Qlik documents Qlik Answers as an assistant built over knowledge bases of unstructured content, "file types such as HTML, DOCX, TXT, and PDF, which are added to knowledge bases and indexed", answering with references back to the source; the workflow runs from preparing content sources through creating an assistant, deploying it and scheduling re-indexing. Qlik Machine Learning is the opposite kind of tool: a code-free experiment over structured, tabular data that trains and ranks several algorithms, deploys the best and then predicts. Question 9 is built on exactly that distinction and it is the single most likely place to lose easy marks — a numeric forecast from years of structured sales history is a prediction problem, not a question for a generative assistant, however capable the assistant is. Insight Advisor is the smallest topic at 10%, and its second objective — "best practices in interacting with Insight Advisor (prompt generation)" — is narrower than it sounds. Qlik’s natural language guidance www.analyticsexam.com Qlik • QAIS Qlik AI Specialist Certification (QAIS) 6 documents three question shapes that work: facts ("What are my sales"), comparisons, which need "vs" or "compare", and rankings, which need "top". It also documents three filter types — time, category and measure — and one hard limit that reads like an exam question waiting to happen: "natural language queries only search the first 100,000 values per field". Synonyms are added through business logic vocabulary. That is a short enough list to learn properly, and at five questions it is worth the hour. The AnalyticsExam QAIS syllabus expands the same five topics and is worth reading alongside this list. SECTION 04 What the Credential Is Worth The honest way to value QAIS is to separate the two halves of it. The Qlik half is worth what any product credential is worth — it signals to an employer already invested in Qlik Cloud that you can build with the AI features they are paying for. The larger, vendor-neutral half is the part with transfer value, because AI literacy is currently the thing every analytics job description asks for and very few candidates can evidence. For the underlying role, the US Bureau of Labor Statistics occupational outlook reports a median annual wage for data scientists of USD 120,230 as of May 2025 , with about 275,600 jobs that year. It projects employment to grow 35 percent from 2025 to 2035 , which it describes as "much faster than the average for all occupations", with roughly 24,800 openings each year over the decade. That is a national median across all experience levels, so it sits below the figures quoted on salary aggregator sites, which skew toward self-reported technology-sector pay. Be clear about what QAIS does and does not do against that backdrop: it is a specialist certification in applying AI, not a data science qualification, and it evidences that you can reason about AI systems and build with Qlik’s, not that you can train a model from scratch. AnalyticsExam’s own write-up of the exam frames the audience as "data analysts, business intelligence developers, and technical professionals" looking to bring AI into their analytics work, and positions the credential as a step toward "roles in data science, AI engineering, business intelligence, and advanced analytics". That is a fair reading of the blueprint. This is a lateral-move credential for someone already in analytics, not an entry point into the field. Where QAIS actually helps Three fifths of the content is vendor-neutral AI literacy that transfers anywhere. It gives a defined syllabus to a subject most people learn in a scattered way. It is the obvious AI credential for an existing Qlik Sense practitioner. The badge is issued through Credly and the badge page states no expiry date. That last point deserves a caveat rather than celebration. A credential with no stated expiry sounds like good value, and administratively it is, but AI is the fastest-moving subject any certification body currently examines. Qlik itself warns on the exam page that content is updated periodically. A QAIS badge earned today will still be valid in three years; whether the knowledge behind it still is depends entirely on you. www.analyticsexam.com Qlik • QAIS Qlik AI Specialist Certification (QAIS) 7 SECTION 05 Getting Certified, Step by Step 1 Fix the vocabulary first AI, ML, deep learning, GenAI, NLP, LLM, training, inference, corpus. 30% of the paper. 2 Learn the LLM lifecycle as a sequence Scope, select and adapt, evaluate, deploy, monitor. Named explicitly in the blueprint. 3 Build one assistant and one experiment A Qlik Answers knowledge base and a Qlik Predict model. The 30% you cannot read up on. 4 Spend an hour on Insight Advisor Facts, comparisons, rankings; the three filter types; business logic synonyms. 10%. 5 Practise against the clock, then book 50 questions, 90 minutes, 73% to pass. Rehearse at 108 seconds a question. Step three is the one people skip and should not. Two of the three product topics are about workflows, and workflows are far easier to recall when you have run them once than when you have only read them. A Qlik Cloud trial is enough: index a handful of PDFs into a knowledge base, create an assistant over it, ask it something and look at how it cites its sources — that alone answers most of what topic 3 can ask, including question 1 in this guide. Then run one machine learning experiment end to end on any tabular dataset and watch the platform train, rank and deploy. An afternoon covers both. On booking, buy and schedule from the Qlik exam-details page rather than through Pearson VUE, for the reason given in Section 02. Qlik states no retake policy on the exam page, so if you need one, ask Qlik Learning before you sit rather than after. When you want to check that recognition holds up under time pressure, AnalyticsExam’s QAIS practice exam is built to this same syllabus, and the Qlik Community education board is where candidates discuss this certification with Qlik staff. SECTION 06 QAIS Sample Questions These come from AnalyticsExam’s QAIS sample questions page, reproduced verbatim and shown here in shuffled order. All ten are single-answer; none asks for more than one response. Answers follow the set. Before you work through them, one observation more useful than the questions themselves. Sorted against the blueprint, this set is three questions from Introduction to AI, four from Business applications and three from Qlik Answers — which tracks the 30/30/15 weighting of those three topics closely. But Qlik Machine Learning and Insight Advisor, 25% of the real paper between them, do not appear at all. So the set is a fair guide to question style and to most of the exam’s weight, and no guide whatever to a quarter of it. Revise to match this sample and you will walk into roughly a dozen questions on AutoML workflows and Insight Advisor prompting whose shape you have never seen. www.analyticsexam.com Qlik • QAIS Qlik AI Specialist Certification (QAIS) 8 The style itself is consistent and worth recognising. Nine of the ten wrap the concept in a short scenario — a named fictional company, one symptom, one decision — and then ask which named concept it illustrates. The options are not bare terms but term-plus-definition pairs, so a distractor is usually a correct definition of the wrong concept. Question 3 is the clearest case: hallucination, underfitting, overfitting and latency are all defined accurately, and only one of them describes the symptom in the stem. Reading the first two words of an option is not enough on this paper. Q1. When a Qlik Answers assistant returns an answer along with references to the specific source documents it drew from, what is the main purpose of including those references? a) To reduce the amount of computing time the underlying model needs to generate each individual response b) To let a reader verify the answer against the original source material and judge its grounding c) To automatically correct any grammatical errors that appear within the generated answer text itself d) To permanently remove the referenced source documents from the knowledge base once they are cited Q2. Shoppers using Bellcastle Retail's new AI shopping assistant abandon the chat when responses take more than a few seconds to appear, even though the answers themselves are accurate and well grounded. Which lever should the team prioritize to address this specific complaint? a) Expanding usefulness by giving the assistant access to a new inventory-lookup tool b) Reducing latency, for example through a smaller model, response caching, or streamed output c) Adapting the data pipeline to add more source documents to the knowledge base d) Increasing the size of the underlying model so that answer quality improves even further than it already has Q3. A generative AI assistant at Driftwood Analytics confidently states a specific statistic that sounds plausible but does not actually appear anywhere in its source material. Which limitation of AI does this best illustrate? a) Hallucination, where the model produces fluent but factually unsupported output b) Underfitting, where a model is too simple to capture real patterns in the data c) Overfitting, where a model memorizes its training data too closely to generalize well d) Latency, where a response takes longer than expected to arrive www.analyticsexam.com Qlik • QAIS Qlik AI Specialist Certification (QAIS) 9 Q4. Within the commonly used nesting of AI-related terms, one subset specifically refers to models built from multi-layer neural networks trained on large amounts of data. Which term does this describe? a) Reinforcement learning, an approach where an agent learns via reward signals from its environment b) Natural language processing, the broad field concerned with machines working with human language c) Deep learning, a subset of machine learning built on multi-layer neural networks d) Retrieval augmented generation, a pattern for grounding generated answers in retrieved content Q5. In a retrieval augmented generation pipeline, a specific step occurs immediately before the language model composes its final response: relevant passages are located within an indexed knowledge source based on the user's question. Which step of the RAG pattern does this describe? a) Retrieval, the step where relevant content is located in the knowledge source before being added to the prompt b) Fine-tuning, the step where model weights are adjusted using labeled examples drawn from the knowledge source over several training passes c) Deployment, the step where a finished application is made available for end users to access d) Tokenization, the general step where text is broken into smaller units before any further processing begins Q6. An AI vendor draws a line between a model's "training" phase and its "inference" phase. Which description best matches the inference phase? a) Collecting and cleaning the raw data that will eventually be used to build a model in the first place b) Permanently deleting a model from a system once it is no longer needed for any further use c) Repeatedly adjusting a model's internal values so its predictions better match already-known outcomes d) Using an already-trained model to produce an output for a brand-new piece of input data Q7. Large language models are typically described as being trained on an extensive collection of text examples gathered for that purpose. What is the common term for this collection of text? a) A ledger, a running record of financial transactions maintained by an accounting system b) A corpus, a large body of text assembled for training or evaluating a language model c) A charter, a formal document that defines the scope and objectives of a business project d) A manifest, a list of files that are included in a single software release package www.analyticsexam.com Qlik • QAIS Qlik AI Specialist Certification (QAIS) 10 Q8. Before writing a single prompt, a team at Ashgrove Consulting spends a week interviewing stakeholders to agree on exactly which business problem a planned generative AI assistant must solve and how success will be measured. Which stage of the LLM Application Project Lifecycle does this activity represent? a) Scoping the use case and defining what success looks like b) Selecting and adapting a base model through prompt engineering, retrieval, or fine-tuning c) Monitoring the deployed application and retraining it as usage patterns drift d) Evaluating the built application for quality, groundedness, and safety before release Q9. A revenue operations team wants to forecast next quarter's sales volume for each product line from several years of historical, structured sales data. Which statement about the fit of Qlik Answers for this task is most accurate? a) Qlik Answers is a poor fit only because it is structurally unable to process numbers of any kind, structured or unstructured b) Qlik Answers is the best fit as long as the historical sales data is first converted into a set of written narrative summaries c) Qlik Answers is a poor fit, since this is a structured-data prediction task better suited to a predictive machine learning approach d) Qlik Answers is the best fit, since generative assistants are generally more accurate than predictive models at any numeric forecasting task Q10. A team has just finished collecting and cleaning a raw dataset of customer records for a new predictive model, and no modeling work has started yet. Which activity typically comes next in the machine learning workflow? a) Deploying the trained model into a live production environment for end users to access immediately b) Monitoring the live model for signs of degrading accuracy over an extended period of time c) Publishing the final evaluation report comparing the chosen model against a business benchmark figure d) Engineering and selecting the features that will be used as model inputs Answer key Q1 Q2 Q3 Q4 Q5 Q6 Q7 Q8 Q9 Q10 b b a c a d b a c d Two of these are worth studying rather than answering. Question 5 walks the RAG pattern one step at a time and asks you to name the step where passages are located in an indexed source before the model writes anything — retrieval, and the distractors are fine-tuning, deployment and tokenisation, each defined correctly. RAG is named explicitly in the Qlik Answers objectives, so knowing the sequence rather than the acronym is the point. And question 10 asks what comes next when a dataset has been collected and cleaned but no modelling has started; the answer is feature engineering, and the three wrong options are all real stages of the same workflow placed at the wrong moment. Order questions of www.analyticsexam.com Qlik • QAIS Qlik AI Specialist Certification (QAIS) 11 that shape are the cheapest marks on this exam and the easiest to lose by reading quickly. SECTION 07 Where to Go Next A word on study material, because this is a subject where the internet is unusually unreliable. General AI content ages in months, and a great deal of what is written about LLM applications describes tooling that did not exist when this blueprint was written or has been renamed since. Work from Qlik’s own exam-details page for the topic list and weightings, use Qlik’s product documentation for the three product topics rather than third-party summaries of it, and keep the wider Qlik catalogue in view on the AnalyticsExam Qlik certification list when you decide whether a Sense Business Analyst or Data Architect credential should follow this one. Quick reference Official exam page, topics and weightings - learning.qlik.com Preparation path and purchase - learning.qlik.com, not Pearson VUE Syllabus, sample questions and practice test - analyticsexam.com/qlik Product documentation for all three Qlik topics - help.qlik.com Badge criteria after you pass - credly.com/org/qlik Before you book — the guide in one card Duration 90 minutes (Qlik; 120 elsewhere) Questions 50 (Qlik) Passing score 73% — 37 of 50 Exam fee USD 250 (AnalyticsExam syllabus) Booking Qlik Learning, not Pearson VUE Badge Credly, no stated expiry Heaviest topics Introduction to AI and Business applications, 30% each Lightest Insight Advisor (10%) Not about Qlik 60% of the paper Objectives 17 across five topics Sample set covers Topics 1, 2 and 3 only Sample set misses Qlik Machine Learning and Insight Advisor Good luck. The useful thing about this paper is that Qlik tells you exactly where the weight sits, and where it sits is not where most candidates expect. Get the AI vocabulary and the LLM lifecycle genuinely solid, spend an afternoon actually building an assistant and a model, and the 73% stops looking like the obstacle it first appears to be.