What is this?
DSPy is a Python framework for programming — rather than prompting — language models: you build modular AI systems in code, and its algorithms automatically optimize the prompts and weights.
DSPy is a Python framework for programming — rather than prompting — language models: you build modular AI systems in code, and its algorithms automatically optimize the prompts and weights.How is that different from just writing prompts myself?You don't hand-tune brittle prompt strings — you write compositional Python code, and DSPy uses that to teach the language model to deliver high-quality outputs.
You don't hand-tune brittle prompt strings — you write compositional Python code, and DSPy uses that to teach the language model to deliver high-quality outputs.Who is it for — do I need to be an ML researcher?Developers building language-model systems in Python, from simple classifiers to sophisticated RAG pipelines and agent loops. Researchers are a core audience too — the framework ships alongside a series of papers.
Developers building language-model systems in Python, from simple classifiers to sophisticated RAG pipelines and agent loops. Researchers are a core audience too — the framework ships alongside a series of papers.What can I build with it?The authors call out simple classifiers, sophisticated RAG pipelines, and agent loops — essentially any modular system that calls a language model.
The authors call out simple classifiers, sophisticated RAG pipelines, and agent loops — essentially any modular system that calls a language model.What does the name DSPy stand for?Declarative Self-improving Python: you declare your system in Python code, and the framework's optimizers improve its prompts and weights on their own.
Declarative Self-improving Python: you declare your system in Python code, and the framework's optimizers improve its prompts and weights on their own.Is there published research behind it?Yes — a whole line of papers, from DSP (Demonstrate-Search-Predict, Dec 2022) through the original DSPy paper (Oct 2023) to GEPA (Jul 2025) on reflective prompt evolution outperforming reinforcement learning.
Yes — a whole line of papers, from DSP (Demonstrate-Search-Predict, Dec 2022) through the original DSPy paper (Oct 2023) to GEPA (Jul 2025) on reflective prompt evolution outperforming reinforcement learning.What are similar projects?SkillOpt, hermes-agent-self-evolution, PocketFlow, rrsi, imp, FinGPT, ai-agent-book, langchain
SkillOpt, hermes-agent-self-evolution, PocketFlow, rrsi, imp, FinGPT, ai-agent-book, langchainWhich topics is it in?llm-framework, prompt-optimization, self-improvement, optimization, fine-tuning, nlp, prompt-engineering, artificial-intelligence, rag, developers, framework, research
llm-framework, prompt-optimization, self-improvement, optimization, fine-tuning, nlp, prompt-engineering, artificial-intelligence, rag, developers, framework, research