DSPy Logo

DSPyWeekly

Your Weekly Dose Of All Things DSPy

Articles and Tutorials
Videos
Every Friday

You'll receive verification email in max 1 minute. Do check SPAM just in case. Unsubscribe anytime.

Subscribed by employees and students at

Apple
Microsoft
LinkedIn
Columbia University
JioStar
Freshworks
Netflix
Databricks
MIT

".@DSPyOSS is so good that i'm kind of sad how many hours i spent struggling without it last year"

Joshua Weaver
Joshua Weaver
Director @txoji Attorney with a background in tech and entrepreneurism.

"Both DSPy and (especially) GEPA are currently severely under hyped in the AI context engineering world."

tobi lutke
Tobi Lutke
CEO by day, Dad in evening, hacker at night.

"We've reached that stage where every single day of the week (and every weekend), there are *several* really cool @DSPyOSS research papers, open-source applications, production use cases, or deep-dive tutorials, etc."

Omar Khattab

DSPyWeekly Issue No #9

Published on October 31, 2025

📚 Articles

DSPy Community Map - Find Other DSPy users near you and add yourself too

Announcing the launch of the new Global DSPy Community Map, an interactive directory designed to visualize and connect our worldwide user base. We invite all DSPy developers and enthusiasts to add themselves to the map by creating a public profile. This is a chance to showcase your expertise, share your bio and social links, and network with fellow users in your region. The map also features a "Looking for work" option, turning it into a valuable resource for discovering talent. To get started and put yourself on the map, simply visit https://dspyweekly.com/community/add/, fill out your details, and submit your profile.

REFRAG implementation in DSPy - By Marcus Johansson

A comprehensive, production-ready framework for Retrieval-Enhanced Fragmented Reasoning and Generation (REFRAG) that revolutionizes how large language models process and reason with retrieved information. Built on DSPy, this enterprise-grade solution provides advanced benchmarking, memory-enhanced capabilities, and sophisticated analysis tools for next-generation AI applications.

A practical comparison of DSPy and LlamaBot for structured LLM applications

In this blog post, I share my hands-on comparison of DSPy and LlamaBot for building structured LLM applications, using a real-world expense extraction example. I explore how each framework handles schema design, type safety, and prompt optimization, highlighting their strengths and trade-offs. Curious which approach might best fit your next LLM project?

Interactive UI for DSPy prompt optimization

Build and test signatures interactively. Create multiple cells to experiment with different prompts and see results side by side.

DSPy - Butter Integration

Butter is a cache that identifies patterns in LLM responses and saves you money by serving responses directly. It's also deterministic, allowing your AI systems to consistently repeat past behaviors.

DSPy creators dissertation that started it all

This culminates in the ColBERT paradigm for neural IR and in the DSPy framework for natural language programming. We demonstrate through these contributions that making broad progress in AI is not restricted to training larger models, but can take the form of designing general tools that grant AI researchers and developers the capacity to controllably improve their systems via composition, to transparently ground their systems’ responses in massive knowledge collections, and to scalably deploy their systems via new algorithms and new compositions of smaller LMs.

BAML and DSPy

The new BAML adapter (which subclasses the existing JSON adapter) avoids using JSON schema for formatting the data model in the prompt. Instead, it uses the format popularized by BAML, and it instantly improves the quality of structured outputs. Using the BAMLAdapter in dspy is incredibly simple - just pass it as a parameter to your LM, and the adapter takes care of the rest.

🎥 Video

DSPy Boston Video Playlist

Big shoutout to Team Weaviate for organising and recording these.

Build Production-Ready LLM Apps with Prompt Optimization | DSPy + Dagster

In this Dagster Deep Dive, Alex Noonan and Colton Padden demonstrate how to escape the "prompt spaghetti" trap that plagues most LLM applications. Learn how combining DSPy's declarative framework with Dagster's orchestration creates maintainable, testable, and automatically optimized AI systems.

🚀 Projects

assagman/dsgo

Unofficial DSPy implementation in Golang | Language: Go | License: MIT License

baserow/udspy

udspy addresses a specific use case: resource-constrained environments. DSPy's dependency on LiteLLM (which requires ~200MB of memory when loaded) makes it challenging to use in contexts with limited resources.| Language: Python | License: MIT License

alibabadoufu/dspy-bench

dspy-bench is an open-source, extensible Python project to evaluate every DSPy strategy & optimizer on user datasets. Language: Python | License: MIT License

jordan-barrett-jm/llm-page-extraction-gepa-dspy

Demonstrates how to use the GEPA prompt optimizer in DSPy to improve a financial statement page extractor. | Language: Jupyter Notebook

💼 Jobs

Open Call for Fellowship Applications, 2026 and 2026-2027 at The Berkman Klein Center for Internet & Society at Harvard University i

Applications are now open for scholars and practitioners who wish to hold a fellowship with the Berkman Klein Center (BKC). We seek candidates who will propose and lead independent research initiatives aligned with BKC’s interdisciplinary AI research agenda. Fellows appointed through this call will bring enthusiasm for working in interdisciplinary and intersectoral environments; fluency in communicating and translating between technical and non-technical stakeholders and audiences; excitement about working with and mentoring students; and a shared commitment to BKC's public interest mission and to open-source, accessible AI research. We strongly encourage fellows to be in residence in Cambridge, MA, although non-resident fellowships will be considered on a case-by-case basis. We welcome applications for two distinct appointment periods: January-August 2026 2026-2027 Academic Year (September 2026 - August 2027) More information about our call for applications is detailed below. Applications will be accepted until Friday, December 5, 2025 at 11:59 p.m. ET. Familiarity with modern agent frameworks (e.g., DSPy) and communication protocols (e.g., MCP, A2A) More information on the website