DSPyWeekly Issue No #4
Published on September 26, 2025
📚 Articles
From Mimicry to Authenticity: Systematic Voice Evolution in DSPy
This article teaches how to develop authentic AI-generated voice using DSPy and genetic-Pareto optimization to systematically explore a multi-dimensional "voice space" rather than mimicking existing writing styles. Instead of training AI to copy someone's voice, the system treats voice as an 8-dimensional vector (covering aspects like authority, weirdness, evidence style, etc.) and uses multi-objective optimization to evolve configurations that balance truthfulness, authenticity, and engagement. The key insight is that genuine AI voice emerges from systematic evolution toward specific goals rather than imitation of existing authentic expressions.
GEPA Advanced Features - Documentation
This page offers an advanced guide for optimizing DSPy programs using dspy.GEPA, focusing on two key customization features. You can create custom instruction proposers to generate new prompts, which is especially useful for handling multi-modal inputs, setting length constraints, adding domain-specific knowledge, or optimizing for specific language models. Additionally, you can implement custom component selectors to control the optimization order, moving beyond the default round-robin method to select program parts based on dependencies or performance. The guide provides detailed protocols and examples, showing how these two features can be combined for highly tailored optimization workflows.
DSPy + OpenRouter: Why the Standard Integration Falls Short (And How to Fix It)
The standard DSPy OpenRouter integration has a critical limitation: it doesn't support model failover and always shows "LiteLLM" as the app name in OpenRouter's leaderboards. Here's how to fix it with custom headers to get automatic failover and proper app attribution.
Raising the bar on Amazon DynamoDB data modeling | AWS Database Blog
Build using DSPy.
In Person DSPy Boston Event: Build Applications with Weaviate, AWS, and DSPy · Luma
Talks - Compound Retrieval Systems with Connor Shorten, Nova Customization with Vikram Shenoy, Arbor with Noah Ziems, DSPy 3.0 with Omar Khattab the creator of DSPy.
DSPy.rb now has OpenRouter integration.
This is a mock summary for the article at https://x.com/highwayvaquero/status/1971314087574618192.
Essential GitHub Repositories for AI Engineers
A collection of GitHub repos for AI engineers.
🎥 Video
Getting The Same Result Every Time With DSPy? Here is the fix
How to disable dspy cache?. I have personally tripped over this initially, @datmt_dev made a video on it.
DSPy Masterclass — 5 Real-World Use Cases for AI Engineers
Nice high quality video teaching DSPy.
🚀 Projects
nielsgl/dspy-profiles
A companion tool for the DSPy framework to manage configuration profiles, inspired by the AWS CLI and Databricks. dspy-profiles allows you to define, switch between, and manage different DSPy configurations for various environments (e.g., development, staging, production) without cluttering your code. | Language: Python | License: MIT License
evalops/orbit-agent
A brutally honest "high‑orbit" startup advisor you can text or run from the CLI. Built with DSPy, it provides opinionated, YC-style advice and financial tools for founders. | Language: Python | Stats: 10 watchers
Scale3-Labs/dspyground
A tool kit for generating high quality prompts using DSPy GEPA optimizer | Language: TypeScript | License: Other
evalops/dspy-micro-agent
Minimal agent runtime built with DSPy modules and a thin Python loop. Includes CLI, FastAPI server, and eval harness with OpenAI/Ollama support. | Language: Python | Stats: 6 forks, 45 watchers
haasonsaas/dspy-0to1-guide
A comprehensive 0-to-1 guide for building self-improving LLM applications with DSPy framework | Language: Python | Stats: 9 forks, 139 watchers
Thanks for reading DSPyWeekly!