Gains:
- Ability to understand the structure of an audit finding (situation, criterion, cause, effect, recommendation) and use artificial intelligence to produce a draft finding and management letter
- Ability to understand auditor opinion types and report language and prepare clear, evidence-based draft texts with artificial intelligence
- Ability to understand that the opinion decision and report content belong to the auditor and that artificial intelligence only accelerates the language and structure
All the effort of the audit is finally collected in one place: the report. Whether it is an independent auditor's opinion (conviction about whether the financial statements are true and fair), an internal audit report, or a management letter (a letter reporting internal control weaknesses and suggestions to management), the value of the audit depends on the reader's ability to understand it and take action. If a good examiner gives a great test and writes it poorly, he loses half the value of his job. This is where AI is a powerful aid—in accelerating language, structure, and consistency. But let's draw the line from the beginning: the opinion, characterization and conclusion in the report are the auditor's; AI only speeds up expression and skeleton.
In this unit we will focus on two things: the anatomy of a solid audit finding and how to draft it with AI; and the types of auditor opinions and the subtleties of report language.
Anatomy of a finding: DKNEÖ
An effective audit finding consists of five elements; you can remember them with an abbreviation — DKNEÖ:
- Situation (what was found): Observed fact, in numbers. "There was no second confirmation for 12 payments; the total amount is 340,000 TL."
- Criterion (what rule was violated): What policy, standard, or control expectation was violated. "The company acquisition procedure requires double approval for payments over 50,000 TL."
- Why (root cause): Why did this happen? “The approval workflow is structured to allow single approval for urgent payments.”
- Impact (consequence/risk): Why is this important? "Risk of unauthorized or erroneous payments; internal control weakness."
- Recommendation (what to do): Concrete, workable solution. "The immediate payment exception should be removed or subject to mandatory second approval later."
This structure makes the finding both convincing and fair: it presents what, according to what rule, why, with what result and how to resolve it. AI is very good at fitting your raw findings into these five headings. But the content of each title must be based on evidence; in particular the "Condition" numbers and the "Criteria" reference must be verified (if the criterion is a standard, the confirmation discipline from the previous unit applies).
Tip: Pay attention to your tone when writing the finding. The audit finding targets the process and control, not the person. Write "inadequate approval control design" rather than "accounting manager careless". Instruct the AI to “use constructive, process-oriented language, not accusatory.”
Types of auditor opinions: weight of language
The heart of the independent auditor's report is the opinion, and there are four basic types. Before printing them to the AI, the auditor decides which view is appropriate; This is a judgment, not a language choice:
Opinion type
when
Meaning
Positive opinion
No major errors
Tables are accurate and truthful
Limited positive (conditional) opinion
Significant but uncommon inaccuracy/limitation
"Except for this" is correct.
negative opinion
Important and common mistake
Tables are not accurate and realistic
Avoid expressing an opinion
Insufficient evidence collected, widespread uncertainty
No comment possible
The choice of view type is the most weighty judgment of the audit and is never left to the AI. After you make the decision, the AI can draft the language of the relevant opinion paragraph in a standard format; But it is the auditor's decision based on evidence as to which opinion to give. Similarly, the content of sections such as points of interest and key audit issues can be entrusted to the auditor, and the language to the AI.
Management letter: constructive and action-oriented
The management letter collects internal control weaknesses and suggestions for improvement that do not affect the audit opinion but that management should be aware of. A good management letter presents each item in the DKNEÖ structure, ranks them according to their importance, and includes realistic suggestions. AI is very efficient at turning your messy field notes into an organized, prioritized management letter — as long as you provide the content and importance.
three mini cases
Case 1 — Strengthening a weak finding. One auditor's raw note was: "Some payments are missing approval, please fix it." This was neither convincing nor feasible. The auditor gave the verified numbers and criteria to AI and printed them in the DKNEÖ structure. Result: "In 12 payments (340,000 TL in total) there was no second approval required by the procedure; why urgent payment exception; impact risk of unauthorized payment; recommendation removal of exception." The same finding has now turned into a text that the management can take action on. The content was the auditor's, the structure was the AI's.
Case 2 — Danger of leaving the opinion to the AI. A team member told AI to “write an audit opinion paragraph” despite a material misstatement it detected; The AI produced positive (clean) opinion text by default. The team member put it in the draft without questioning. The responsible person reminded that the detected error requires at least a qualified opinion. Lesson: opinion type is an evidence-based auditor's decision; The AI's default optimistic text is never considered automatic.
Case 3 — Exaggerated/accusatory language. An auditor told the AI to “write hard” on a finding; AI produced harsh, unsupported by evidence statements such as "serious abuse" and "unacceptable negligence". This language was both unfair and legally risky. The auditor softened the text and linked it to evidence and process. Lesson: report language should be proportionate to the weight of the evidence; The auditor balances the dramatic language of the AI.
Weak prompt / Strong prompt
Weak prompt:
Write a report for this audit and give your opinion.
Problem: no content, opinion decision left to AI. AI produces either a blank template or a made-up “opinion” without evidence.
Powerful prompt:
Your role: you are a reporting assistant to an independent auditor. I decide the type of opinion and the content of the findings; You will organize the structure and language. Do not add any findings, numbers or results that do not exist in my content. My raw notes of the finding (verified): - Status: There is no second approval in 12 supplier payments (340,000 TL in total). - Criteria: Company purchasing procedure art. - internal control design weakness. - Recommendation: The emergency exception should be removed or it should be subject to mandatory second approval later. Task: 1) Write this in the DKNEÖ (Situation-Criteria-Cause-Effect-Suggestion) structure, in a constructive and PROCESS-oriented (not accusing person) language, appropriate to the management letter. 2) DO NOT USE harsh expressions (misconduct, intent) that are not supported by evidence; Don't imply cheating unless I say so.3) Suggest "medium" as the level of importance, but I will give the final rating; Write your reason in one sentence.
This demand is strong because the auditor provides the content, requires DKNEÖ structure, imposes process-oriented and measured language, leaves the opinion/importance decision to the auditor, and prohibits fabrication.
Common mistakes
- Leaving the type of vision to the AI. Reducing evidence-based auditor judgment to a matter of language.
- Requesting a report without content. Telling the AI to "write a report" without giving raw findings; produces false findings.
- Exaggerated/accusatory language. Heavy statements beyond evidence; injustice and legal risk.
- Incompletely structuring the finding. Saying "there is a problem" without criteria or impact; It wouldn't be convincing.
- Ignoring importance. Presenting each finding with the same weight and distracting the reader's focus.
Attention: The report is an official document that goes to the public or management. Do not send the AI draft as-is; filter each sentence with evidence, tone, and professional judgment. You stand behind every statement in the report.
In summary
Reporting is the showcase of the audit, and AI language powerfully accelerates structure and consistency: organizes scattered notes into the DKNEÖ structure, prioritizes the management letter, drafts standard opinion paragraphs. But the type of auditor's opinion, the significance of the findings, and the support of each statement with evidence are up to the auditor. The auditor counterbalances the AI's default optimistic or dramatic language; The controller provides the content. The signatory auditor stands behind every sentence in the report.
Application task
Write raw notes of a confirmed (or hypothetical) finding under the five headings of the DKNEÖ. With the powerful prompt pattern above, have the AI edit it in management letter language; Set a process-oriented tone and a "no fabrication" rule. Then read the text and check (1) whether each topic is based on evidence, (2) the language is measured, and (3) whether the suggestion is actionable. Also write a scenario that requires one of the four types of views and justify in a paragraph why you chose that view.
checklist
- [ ] I established the finding in the DKNEÖ (Situation-Criteria-Cause-Effect-Suggestion) structure.
- [ ] "Status" numbers and "Criteria" reference verified; depends on evidence.
- [ ] Language is process-oriented, measured; it doesn't incriminate the person, it doesn't transcend the evidence.
- [ ] The proposal is concrete and applicable.
- [ ] I decide the type of opinion and its importance; I didn't leave it to AI.
- [ ] I balanced the AI's default optimistic/dramatic language with my control judgment.
- [ ] I read the final text line by line; I stand behind every statement.