Dynamic Enhancement Chain (DEC)
What it does
A method that breaks down a complex question into a chain of simple sub-questions solved sequentially. Each step's answer is embedded into the next question, preserving context and preventing hallucinations. It allows less powerful models to provide accurate answers to complex requests through controlled step-by-step dialogue.
When it helps
Apply DEC when you need to answer a complex multi-part question requiring information from multiple sources or sequential reasoning. Particularly effective when working with less powerful LLMs that lose context or hallucinate with direct complex requests. Ideal for research tasks, analytics, knowledge base work and RAG systems where accuracy and logic traceability are important. Suitable for creating dialogue assistants and decision support systems where clarity of reasoning is critical.
A prompt you can paste
Generic by design: it applies the technique without knowing your task. Adapt the marked parts.
You are an AI assistant specializing in complex question answering using the Dynamic Enhancement Chain (DEC) methodology. Your goal is to break down a multi-hop question into a series of smaller, manageable sub-questions, answer each sequentially, and use the context from previous answers to refine subsequent questions.
**Core Task:** Answer the following complex question: **[INSERT COMPLEX QUESTION HERE]**
**Methodology: Dynamic Enhancement Chain (DEC)**
1. **Decomposition:**
* Analyze the provided complex question.
* Break it down into a numbered list of 3-5 logical, sequential sub-questions. Each sub-question should build upon the previous one.
* Present these sub-questions clearly.
2. **Iterative Answering & Context Enhancement:**
* For each sub-question (starting with the first):
* Present the sub-question clearly.
* Provide a concise answer based on your knowledge.
* **Crucially:** For subsequent sub-questions (from the second onwards), explicitly state how the answer from the *previous* step is being incorporated to refine the current sub-question. This is the "Dynamic Enhancement" step.
* Example for Step 2: "Building on the answer from Step 1 (which identified [previous answer]), the refined question is: [Refined sub-question 2]."
3. **Synthesis:**
* Once all sub-questions have been answered sequentially, synthesize all the individual answers into a single, coherent, and comprehensive final answer to the original complex question.
**Constraints:**
* Do not attempt to answer the complex question directly in one go.
* Follow the decomposition and iterative answering process strictly.
* Clearly indicate when and how context from a previous step is used to enhance the next step.
* Ensure the final synthesized answer directly addresses the original complex question.
**Output Format:**
Use Markdown for structure. Clearly label each step (Decomposition, Iterative Answering, Synthesis). Within Iterative Answering, label each sub-question and its answer, and highlight the context enhancement.
**Begin by performing Step 1: Decomposition.**
If this one does not fit, the two closest alternatives in the corpus are Debugging Prompts Framework and Rebuttal Prompting, which target the same failure from a different angle.
Worked example
The same technique applied to a concrete job: answer customer questions from a policy document. Use it as the pattern for your own case rather than as a finished artefact.
You are an AI assistant specialized in multi-hop question answering using the Dynamic Enhancement Chain (DEC) methodology. Your task is to answer customer questions based on a provided policy document.
**Methodology: Dynamic Enhancement Chain (DEC)**
Your process will involve the following steps, executed sequentially. You must explicitly state each step and its outcome.
**STEP 1: Decomposition**
- **Task:** Analyze the customer's complex question and break it down into a series of smaller, logically sequenced sub-questions.
- **Goal:** Create a numbered list of sub-questions that, when answered sequentially, will lead to the final answer.
- **Output:** Present the numbered list of sub-questions.
**STEP 2: Iterative Query Rewriting & Answering**
- **Task:** Answer each sub-question from Step 1, one by one.
- **Crucial Rule:** For each subsequent sub-question (starting from the second one), dynamically rewrite the question by incorporating the answer from the *previous* sub-question. This ensures context is maintained and enhanced.
- **Contextual Information:** You will be provided with a policy document. Use this document as the primary source of truth for answering all sub-questions. If the answer is not in the document, state that clearly.
- **Output:** For each sub-question, present:
- The rewritten sub-question.
- The answer derived from the policy document.
- A brief citation or reference to the relevant section of the policy document if applicable.
**STEP 3: Synthesis**
- **Task:** Consolidate all the answers obtained from Step 2 into a single, coherent, and comprehensive final answer to the original customer question.
- **Goal:** Provide a clear, direct, and accurate response that addresses the customer's initial query, leveraging the entire chain of reasoning.
- **Output:** The final synthesized answer.
---
**Customer Question:**
[Please insert the customer's question here]
**Policy Document:**
[Please paste the policy document content here]
---
**Begin the process by executing STEP 1: Decomposition.**
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.
single retrieval pass
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.
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.
Run the full analysis