6 techniques drawn from the 261 in our corpus tagged to this work,
each with the paper it came from. Across the techniques below the recurring targets are answers that are close but wrong, the same prompt giving different answers, invented facts and citations.
The techniques
01
Could you be wrong
Could you be wrong: Debiasing LLMs using a metacognitive prompt for improving human decision making, arXiv:2507.10124
A metacognitive prompt for eliminating LLM bias through two-stage dialogue. After receiving the first answer, the question 'Could you be wrong?' is asked, which forces the model to activate critical thinking and provide counterarguments, risks, and alternative viewpoints that were hidden in the original answer.
Use it whenApply this technique when you need to make important decisions based on LLM answers, verify critical information, or obtain objective analysis. Particularly useful when dealing with controversial topics, risk assessment, business planning, and research, where a one-sided view can lead to erroneous conclusions. Use the method when the model's first answer seems too confident or oversimplified, when you need to identify hidden risks and alternative viewpoints, or when the cost of error is high.
Prompt
### Role
You are an AI assistant tasked with helping an entrepreneur turn a vague goal into an ordered, actionable plan with clear checkpoints. You will employ the "Could You Be Wrong" debiasing technique to ensure the plan is robust and considers potential pitfalls.
### Context
The entrepreneur has a vague goal: "Improve customer retention for our subscription box service." They need a structured plan to achieve this, including specific steps, metrics, and potential risks.
### Task: Develop a Robust Action Plan
**Phase 1: Initial Goal Clarification & Brainstorming**
1. **Initial Goal Statement:** "Improve customer retention for our subscription box service."
2. **Brainstorm Potential Levers:** Identify 5-7 broad areas that could impact retention (e.g., product quality, unboxing experience, customer support, pricing, community building, personalization, communication).
3. **Define Success Metrics:** For each lever, suggest 1-2 key metrics that would indicate improvement (e.g., churn rate, repeat purchase rate, customer lifetime value, Net Promoter Score).
**Phase 2: Debiasing and Risk Assessment ("Could You Be Wrong?")**
Now, critically evaluate the initial brainstorming and proposed metrics. For each identified lever and proposed metric, answer the following:
* **"Could you be wrong about this lever being the most impactful?"**
* Identify potential counterarguments or reasons why this lever might *not* be the primary driver of retention.
* What are the hidden assumptions in focusing on this lever?
* Are there alternative interpretations of the data that would suggest a different focus?
* **"Could you be wrong about this metric accurately measuring success?"**
* What are the limitations of this metric?
* Could this metric be manipulated or misleading?
* Are there external factors that could influence this metric unrelated to our efforts?
* **"What are the most significant risks or overlooked challenges associated with implementing strategies for this lever?"**
* Consider operational, financial, or customer-facing risks.
* Identify potential negative consequences or unintended side effects.
**Phase 3: Prioritization and Action Planning**
1. **Synthesize Debiasing Insights:** Based on the "Could You Be Wrong?" analysis, re-evaluate the initial brainstorming.
2. **Prioritize Levers:** Rank the levers from most to least impactful, considering both potential benefits and identified risks. Focus on the top 2-3 levers.
3. **Develop Actionable Steps:** For each prioritized lever, outline 3-5 specific, concrete actions.
* Assign owners (if applicable, or specify "team").
* Set realistic timelines (e.g., "Week 1-2", "Month 1", "Ongoing").
* Define clear checkpoints or milestones for each action.
4. **Refine Metrics:** Adjust or add metrics based on the debiasing phase to ensure they are robust and meaningful.
**Phase 4: Checkpoints and Iteration**
1. **Establish Review Cadence:** Propose a schedule for reviewing progress and metrics (e.g., weekly team sync, monthly stakeholder review).
2. **Define Iteration Triggers:** What conditions would prompt a re-evaluation or pivot of the plan? (e.g., significant increase in churn, competitor action, new customer feedback trend).
**Output Format:**
Present the plan using clear headings for each phase. Use bullet points for lists and numbered steps. Bold key terms and action items. Conclude with a summary table of the top 3 prioritized levers, their key actions, primary metrics, and major risks.
MK2 at PBIG Competition: A Prompt Generation Solution, arXiv:2507.08335
A method of transforming a prompt into a detailed technical specification with an expert role, evaluation criteria, step-by-step algorithm, and self-checking mechanism. One LLM iteratively creates and improves prompts for another LLM, ensuring depth and answer quality without model retraining.
Use it whenUse MK2 when you need deep, expert results from LLM on complex tasks — business idea generation, strategic planning, technical documentation analysis. Especially effective for tasks where simple prompts give surface-level answers. Suitable for competitions and projects where quality matters more than speed, and there's an opportunity to set up an iterative prompt optimization process. Requires technical expertise to create a multi-component prompt architecture with roles, criteria
Prompt
# КОНТЕКСТ
Предприниматель имеет расплывчатую цель и нуждается в превращении ее в упорядоченный план с контрольными точками.
# ЗАПРОС
Ты — опытный бизнес-стратег и аналитик продуктовых инноваций, специализирующийся на превращении общих идей в конкретные, измеримые планы действий.
Твоя миссия — разработать детальный план для достижения цели, заданной пользователем. План должен быть структурирован, иметь четкие этапы, критерии успеха для каждого этапа и механизм самопроверки.
### Твой план должен быть превосходным по следующим критериям:
1. **Ясность цели:** Цель должна быть четко определена и понятна.
2. **Измеримость:** Каждый этап должен иметь конкретные, измеримые показатели успеха.
3. **Достижимость:** Этапы должны быть реалистичными и достижимыми для предпринимателя.
4. **Релевантность:** Все действия должны прямо способствовать достижению конечной цели.
5. **Структурная целостность:** План должен быть логичным, последовательным и легко читаемым.
### Процесс создания плана (MK2 Methodology):
1. **Анализ цели:** Внимательно изучи предоставленную пользователем цель. Если она расплывчата, задай уточняющие вопросы (или предположи наиболее вероятную конкретизацию, если вопросы невозможны).
2. **Декомпозиция на этапы:** Разбей достижение цели на 4-6 ключевых этапов. Каждый этап должен представлять собой значимый шаг вперед.
3. **Определение метрик успеха:** Для каждого этапа определи 2-3 конкретных, измеримых метрики, которые покажут, что этап успешно завершен.
4. **Формулировка действий:** Для каждого этапа опиши конкретные действия, которые необходимо предпринять для достижения его метрик.
5. **Формулировка питча:** Кратко опиши конечный продукт/результат, который будет достигнут после выполнения всего плана.
### Формат вывода:
Предоставь ответ в виде JSON-объекта со следующей структурой:
PersonaFlow: Designing LLM-Simulated Expert Perspectives for Enhanced Research Ideation, arXiv:2409.12538
PersonaFlow is a prompting technique where LLM simulates multiple expert roles simultaneously to analyze one task. Instead of addressing a universal assistant, a virtual team of specialists from different fields is created, each evaluating an idea from their professional perspective, ensuring more creative and in-depth results.
Use it whenUse PersonaFlow when you need a comprehensive evaluation of an idea, project, or solution from different expert perspectives. Especially effective when developing new products, business ideas, research concepts, or strategic plans where multiple aspects must be considered. Suitable for situations where you lack access to a real interdisciplinary expert team but need deep and multifaceted feedback. The method helps avoid one-sided views and stimulates critical thinking by simulating discussion between specialists from different fields.
Contextualizing Recommendation Explanations with LLMs
Contextualizing Recommendation Explanations with LLMs: A User Study, arXiv:2501.12152
An LLM explanation generation technique that requires explicit connection of a recommendation with the provided user context. Instead of generic descriptions, the model creates personalized justifications by referencing a specific person's history, preferences, or experience, significantly increasing trust and persuasiveness of the response.
Use it whenApply this technique when you need to create personalized recommendations or explanations that build user trust. Especially effective in recommendation systems, client consulting, creating personalized content, or educational materials. Use when you have data about user preferences, history, or context, and you need to not just give advice but persuasively justify it. The method increases transparency of AI decisions and strengthens user engagement.
A Review of LLM-Assisted Ideation, arXiv:2503.00946
A structured three-phase approach to idea generation using LLM. Includes preparation with clear task definition and context, a divergent phase of mass generation of diverse ideas, and a convergent phase of selection and refinement of the best solutions. Replaces single requests with a managed multi-stage process.
Use it whenUse this technique when you need to generate high-quality, well-developed ideas for business, product, or content. Especially effective for complex creative tasks where a simple LLM request produces superficial results. Apply when developing strategies, planning campaigns, creating product concepts, writing content. The method is suitable for situations where quality, depth, and alignment with specific project criteria and constraints matter, not just the quantity of ideas.
Don't Just Translate, Agitate: Using Large Language Models as Devil's Advocates for AI Explanations, arXiv:2504.12424
This technique transforms LLM from a passive assistant into an active critic through special prompts. Instead of generating ideas, the model is tasked with finding weak points, risks, contradictions, and alternative interpretations in a proposed solution. This reduces the risk of over-reliance and helps make more balanced decisions.
Use it whenUse this technique when you need to make an important decision and avoid blind trust in AI conclusions. Especially useful for evaluating business ideas, strategic plans, investment decisions, risk analysis, or hypothesis testing. Apply when you need a multi-sided view of a problem, identification of hidden risks and alternative data interpretations. The method is effective for combating confirmation bias and improving critical thinking quality when working with LLM.
A missing brief. No structure around a request invents the context the model
does not have: your constraints, your audience, your prior decisions. Technique work pays off
after the brief is right, not instead of it.
Stale tricks. Persona lines, politeness and offers of a reward have been
measured repeatedly through 2025 and 2026 and come out close to noise. Anything selling you
those is selling 2023.
Get this written for your actual task
Paste what you are trying to do and the corpus will be matched against it directly.
Free, no account, about ten seconds.
Free · no signup · ~10s
0.00match confidence single retrieval pass
Prompt for your task
That number is low on purpose, and it is real. It is the raw similarity
of one retrieval pass: no specialist read the paper, no judge compared anything, the
first plausible match won.
6,235techniques in the corpus one of which is this page
Picking the right one for a specific task is the work, and it is the work GetDecision does.
This page
one technique, generic prompt
What you just ran
one technique matched to your wording, nothing verified
Full run
ten specialists read the papers in full, a judge ranks the top three for your task and shows its reasoning, generation on the model you pick, saved to your history
See the top three for your taskTen specialists, a judge, and the reasoning shown. Free account, first run included.