CTFL to CT-GenAI: what you already know and what is new

You hold the ISTQB® Foundation Level certification (CTFL), earned last year or ten years ago. Since then, generative AI has crept into your daily testing work: an assistant to rephrase a defect report, a chatbot to suggest test cases, a plugin in your IDE. The CT-GenAI specialist module (Testing with Generative AI) looks like the natural next step. But how much of what you already know will you reuse, and what will be genuinely new?

This article bridges the two programs: what CTFL gives you, what CT-GenAI adds, a mapping table of concepts, and a role-by-role view.

CTFL is the prerequisite for CT-GenAI

It is stated plainly in the CT-GenAI syllabus v1.1 (section 0.6, "Entry Requirement Note") and on the ISTQB® website: you must hold the Foundation Level certificate before taking the CT-GenAI exam. If you do not have it yet, start there; our ISTQB® Foundation certification guide covers the format and how to prepare.

The prerequisite is not a formality. The CT-GenAI syllabus leans on CTFL throughout:

• It uses the terminology of the ISTQB® Glossary (section 0.9).
• It refers explicitly to the CTFL v4.0 syllabus, for example to define test design (section 2.2.2), for the test planning exercises (estimation, prioritization) and for evaluating tools when adopting generative AI.
• The "classic" keywords of chapter 2 are pure CTFL vocabulary: acceptance criteria, test case, test condition, test data, test design, test report, test script.

One practical point: if your CTFL is from version 3.1, revisit the v4.0 concepts, since that is the version the CT-GenAI syllabus cites.

What CTFL gives you

A shared vocabulary

Test basis, test condition, coverage, test oracle, regression, monitoring and control: CT-GenAI exam questions use these terms without redefining them. A candidate who confuses a test condition with a test case loses points on questions that appear to be about prompting.

The test process

Section 2.2 of the CT-GenAI syllabus is built on the test activities you learned in CTFL: test analysis, test design and implementation, automated regression testing, test monitoring and control. For each activity, the question becomes: which tasks can generative AI support, and with which prompting technique?

Test techniques

Equivalence partitioning, boundary value analysis, risk-based prioritization: the CT-GenAI syllabus mentions these as things AI can suggest or apply. It does not teach them again. It is up to you to judge whether the LLM applied them correctly.

The review reflex

CTFL taught you that reviewing a work product finds defects early. CT-GenAI applies that reflex to LLM output: cross-verification against the test basis, logical validation, running the generated tests. It is probably the most useful CTFL skill in the whole module.

What is genuinely new in CT-GenAI

Chapter	What you learn	Example concepts
1. Introduction to generative AI	How an LLM works and what it can do for testing	Tokenization, context window, embedding, foundation, instruction-tuned and reasoning LLMs, multimodal model
2. Prompt engineering	Writing, choosing and improving prompts for test tasks	Six-part prompt, prompt chaining, few-shot, meta prompting, system prompt, evaluation metrics
3. Risks	Recognizing and mitigating what goes wrong	Hallucinations, reasoning errors, biases, non-determinism, temperature, privacy, Context Manipulation, energy, regulations
4. Infrastructure	Understanding LLM-powered test tools	Front-end/back-end architecture, RAG, vector database, agents, fine-tuning, LLMOps
5. Adoption	Bringing generative AI into a test organization	Shadow AI, strategy, LLM/SLM selection, adoption phases, skills, changing roles

The deepest change is not technical. In CTFL, a test tool produces predictable results. With an LLM, the same input can produce two different outputs, and a wrong output can be beautifully written. CT-GenAI teaches you to work with a collaborator that is confidently wrong at times.

Mapping table: CTFL concepts reused in CT-GenAI

CTFL concept	Where it comes back in CT-GenAI	What changes with generative AI
Defects in the test basis, reviews	1.2.1, 2.2.1 (LO 2.2.1)	The LLM flags ambiguities and inconsistencies; you decide
Test conditions and coverage	2.2.1	The LLM generates conditions and a coverage analysis you must check
Risk-based prioritization	2.2.1, 2.2.2	The LLM proposes priorities; watch for reasoning errors (3.1.1)
Equivalence partitioning, boundary values	1.2.1, 2.2.1	The LLM suggests the technique and values; you check the boundaries
Test cases, procedures, scripts	2.2.2 (LO 2.2.2)	Draft testware generation, including Gherkin style
Expected results	1.2.1	Test oracle generation, to be cross-checked with the test basis
Test data	2.2.2	Privacy-preserving synthetic test data
Regression testing, test automation	2.2.3 (LO 2.2.3)	Keyword-driven scripts, impact analysis, self-healing tests
Test monitoring, control, completion	2.2.4 (LO 2.2.4)	Generated summaries, dashboards and completion reports
Defect reports	1.2.1, 2.2.3	Assisted writing and enrichment
Test planning and estimation	3.1.1, hands-on objective HO-3.1.2b	Tasks where LLMs make reasoning errors
Test tools	5.1.3, 5.1.4	Selecting and evaluating LLMs/SLMs, adoption phases

Example: when CTFL catches the LLM

Take a fictional travel insurance user story: "The online quote is offered to travelers aged 18 to 75 inclusive, for a trip of 1 to 90 days. Beyond 90 days, the quote is replaced by a contact form." We ask the LLM:

Role: you are a test analyst.
Context: travel insurance quote application, system testing.
Instruction: identify the equivalence partitions for age and trip length,
then apply 2-value boundary value analysis.
Input data: user story US-88 above.
Constraints: use only the rules stated in the user story.
Output format: table ID | Parameter | Value | Expected result.

Output excerpt (illustrative):

TC-01 | Age     | 17       | Quote refused
TC-02 | Age     | 18       | Quote offered
TC-03 | Age     | 75       | Quote refused
TC-04 | Age     | 76       | Quote refused
TC-05 | Length  | 1 day    | Quote offered
TC-06 | Length  | 90 days  | Quote offered
TC-07 | Length  | 91 days  | Error message "invalid trip length"

Two defects, which a trained tester spots quickly:

• TC-03: 75 is the inclusive upper boundary of the valid partition. The expected result is wrong, and TC-03 and TC-04 now test the same side of the boundary. CTFL boundary value analysis reveals it; CT-GenAI calls it a reasoning error (a misread conditional rule).
• TC-07: the user story specifies a contact form, not an error message. The expected result does not come from the test basis: it is a hallucination in the syllabus sense.

CTFL gives you the means to detect the problem; CT-GenAI gives you the vocabulary to name it and the techniques to prevent it (sharper constraints, prompt chaining with human verification, cross-verification). For the prompt structure used here, see our article on the structured prompt for testers.

Who CT-GenAI is for, by role

The syllabus targets anyone using generative AI for testing: testers, test analysts, test automation engineers, test managers, user acceptance testers and developers. It also lists project managers, quality managers, business analysts, IT directors and consultants who want a basic understanding.

Role	What you will use most	Key chapters
Tester, test analyst	Prompts for analysis and design, spotting LLM errors	2, 3
Test automation engineer	Keyword-driven scripts, self-healing tests, agents, RAG	2.2.3, 4
Test manager, QA lead	Assisted monitoring and control, privacy, strategy, model selection, change management	2.2.4, 3.2, 5
Developer, business analyst, project manager	What generative AI changes in testing, and its limits	1, 3, 5

If your job is to test a product that contains AI, the question is a different one: see our comparison CT-GenAI vs CT-AI.

What to remember for the CT-GenAI exam

• CTFL vocabulary is assumed. The chapter 2 keywords (test condition, acceptance criteria…) are examinable, and the scenarios are full of them.
• The chapter 2 K3 questions are test process questions. LO 2.2.1 to 2.2.4 cover test analysis, test design and implementation, regression testing, and monitoring and control. To pick the right answer, you need to know what these activities produce.
• LO 2.2.5 (K3): select the prompting technique that fits the context and the test task. The task is described in CTFL terms.
• LO 3.1.2 (K3): identify hallucinations, reasoning errors and biases. The example above is a textbook case: without boundary value skills, the error slips through.
• A question can draw on several syllabus sections, and therefore on several layers of knowledge.

The full program, chapter by chapter, is in our CT-GenAI certification guide.

FAQ

Can I take CT-GenAI right after CTFL?

Yes, as soon as you hold the CTFL certificate. It is actually a good time: the vocabulary and test process are fresh. You will just need more time on chapters 1 and 2 if you have never practiced prompting.

Do I need to know how to code?

The only prerequisite is CTFL. The objectives are about using generative AI in test activities, not about programming. The automation topics (scripts, agents) require understanding the concepts, not writing code.

Does CT-GenAI replace an Advanced Level module?

No. It is a specialist module. The syllabus presents the Core Advanced Levels (Test Analyst, Technical Test Analyst, Test Manager, Test Engineering) as possible next steps, not as equivalents.

---

You have your CTFL and want to know where you stand on the rest? The free diagnostic takes 5 minutes and 10 questions, and shows which CT-GenAI chapters need the most work in your case.

Sources

• ISTQB®, CT-GenAI certification page: CTFL prerequisite, target audience, documents.
• ISTQB®, CTFL v4.0 certification page: v4.0.1 syllabus chapters, role as a prerequisite.
• ISTQB®, Certified Tester Specialist Level Syllabus – Testing with Generative AI, v1.1, 2026-04-27, sections 0.2, 0.3, 0.6, 0.9, 2.2 and 3.1.
• ISTQB®, Certified Tester Foundation Level Syllabus, v4.0.1, 2024.