Uncle Sunny Academy
📄 PDF 🇬🇧 🇭🇺 · 🎧 Audiobook 🇬🇧 🇭🇺 · ✉ delivered by email
How one person and four artificial intelligences work together so that mistakes surface before release, not after. Educational content — not trading, not investment.
SOL
Structure AI
VERA
Review AI
FABER
Build AI
ARGUS
Observer AI
What will you learn here?
On this page you can read the fundamentals for free: the five prompt elements, the most common mistakes, and three exercises. The full publication goes further — it shows how to coordinate several AI systems so that mistakes surface before release. Twenty-eight pages in Hungarian, twenty-seven in English, with eleven diagrams, from the log of a live project.
Free
The five prompt elements
Effective AI communication is not a talent but a learnable structure. This section comes from the second chapter of the publication, and it is free to read in full.
Goal
What you want to achieve — stated precisely and measurably. Not 'write something', but 'write a 3-sentence summary for audience X in format Y'.
Bad: 'Explain prompt engineering.' Good: 'Write a 5-sentence, non-technical explanation of prompt engineering for a high school student, without bullet points.'
Context
What background information the AI receives — what it knows, what it does not. Context is always explicit, never assumed.
Bad: 'Continue the previous work.' Good: 'In the previous session we worked on topic X, the result was Y. Now we continue in direction Z.'
Constraints
What the AI must NOT do — exclusions, prohibited content, scope boundaries. Constraints are not distrust — they are precision.
Bad: (no constraints given). Good: 'Do not give financial advice. Do not use jargon. Do not exceed 200 words.'
Format
What format you want the answer in — list, table, prose, code, JSON. Format determines usability.
Bad: 'Give a summary.' Good: 'Give a 3-column table: Topic | Key point | Next step.'
Triple-Check Row
Ask the AI to verify its own answer from three angles: (1) does it meet the goal, (2) does it respect the constraints, (3) does it contain any assumption you have not confirmed.
Final prompt line: 'Before answering, verify: (1) do you meet the goal, (2) do you respect the constraints, (3) does your answer contain any assumption?'
Common Mistakes
| Mistake | Consequence | Fix |
|---|---|---|
| Vague goal | The AI generalises, the answer is unusable | Add measurability: who, what, in what format, what length |
| Implicit context | The AI assumes — and the assumption is usually wrong | Provide all relevant background information explicitly |
| No constraints | The AI exceeds its scope, unwanted content may appear | Always specify what the AI must NOT do |
| No triple-check | Errors pass unnoticed into the final output | For every critical output, request the AI's self-verification |
Practice Exercises
Write a 5-element prompt on a topic of your choice. Check: are all 5 elements present?
Take a previous prompt that did not get a good answer. Identify which element was missing.
Write a prompt whose last line is the triple-check request. Compare the answer with a version without triple-check.
Available now
Verified AI Workflow
The full publication shows in fourteen chapters how several AI systems can be coordinated over months so that mistakes surface before release: enforcing evidence, the two kinds of task, the stop condition, and what the word "done" actually means. Five case studies from a single working day — including one fix that nearly broke a working system. Every rule was born from a specific failure.
Core Principles
Every step has a measurable success criterion
No 'good faith' — only evidence
Every AI output goes through triple verification
Context is always explicit, never assumed
Errors must not be hidden — they must be documented immediately
The full publication
One-time purchase, no subscription. You receive both editions — in English and Hungarian, as PDF, by email. The methodology keeps developing: new editions of this publication released up to 31 December 2027 will be sent free of charge to the email address used at purchase.
Educational content. Not investment advice.
Uncle Sunny Academy contains educational and informational content. The methodology presented is illustrative in nature. This is not investment, financial, legal, or medical advice. Data Analytic Investments assumes no responsibility for the results of applying the methodology presented here.