As prompt engineering continues to evolve, a new paradigm is emerging that could outshine familiar strategies like Chain-of-Thought (CoT) and ReAct. Introducing Chain of Draft (CoD) β a multi-step prompting technique that turns large language models into more deliberate, thoughtful writers.
Chain of Draft is a prompting method where the AI model is instructed to generate multiple drafts of a response, reflect on them, and then produce a refined, high-quality final answer.
Rather than immediately jumping to the final output, CoD encourages the model to simulate a thoughtful writing process β just like how humans draft, review, and revise content.
Technique | Focus | Strength | Limitation |
---|---|---|---|
Chain of Thought (CoT) | Step-by-step reasoning | Good for logic/math | May miss nuance or creativity |
ReAct | Reasoning + action (e.g., tools) | Interacts with environments | Complex, needs plugins |
Chain of Draft (CoD) | Drafting & self-reflection | Versatile, thoughtful output | Longer inference time |
Big ideas begin with small steps.
Whether you're exploring options or ready to build, we're here to help.
Letβs connect and create something great together.