DPS Labs is active in the infrastructure and tooling layers of the AI agent stack. Their primary contribution is the development of documentation pipelines and context management tools that facilitate the training and deployment of specialized agents. By leveraging Arweave's permanent storage, DPS Labs provides a foundation for 'verifiable' AI, where the data used to inform an agent's decisions is decentralized and immutable.
The studio's 'LLM Fuel' and 'Pocket Prompt Suite' are direct responses to the needs of developers building agentic workflows. These tools handle the 'rag-to-agent' pipeline, ensuring that Large Language Models have access to clean, aggregated context. For people building or using agents, DPS Labs matters as a provider of the 'glue' that connects autonomous software to reliable data sources and organized human intent.
DPS Labs is the engineering studio of Dylan Shade, a software engineer who has navigated between enterprise systems at JPMorgan Chase and the decentralized frontiers of the Arweave network. Since 2019, the studio has released a series of products ranging from music theory applications to specialized documentation pipelines for large language models. The central theme of the current work at DPS Labs is the management of information context, specifically how developers and AI agents interact with high-volume documentation and codebase data.
The studio is a heavy contributor to the Arweave 'permaweb' ecosystem. This work is primarily focused on the AO protocol, a decentralized computing layer built on Arweave. By building tools like LLM Fuel, DPS Labs addresses the data ingestion bottleneck for AI models. LLM Fuel is a documentation aggregation pipeline designed to feed AI training processes, ensuring that agents have access to verified, permanently stored documentation. This is a practical solution to the 'hallucination' problem in developer-facing AI, where models often cite outdated or incorrect API documentation.
Beyond data pipelines, DPS Labs maintains the Pocket Prompt Suite. This is a set of interfaces — CLI, TUI, and a Raycast extension — designed for context management. As developers increasingly rely on LLMs to assist in coding, the bottleneck shifts from writing code to managing the context of the prompts sent to the model. Pocket Prompt Suite is an attempt to centralize this workflow locally, moving prompt engineering away from fragmented browser tabs and into the developer's primary workspace.
This focus on developer experience is also evident in Shade's contributions to OpenCode, where he developed jj-opencode, a workflow plugin that enforces a 'define-before-implement' pattern. This methodology is particularly relevant for agentic development, where clear definitions and specifications are required for autonomous systems to execute tasks effectively.
Dylan Shade founded DPS Labs in 2019 while studying at the University of Kentucky. The studio's consumer-facing applications, such as Vivace Theory and Chord Solver, have served over 25,000 users. Shade’s background as a 2x hackathon champion at JPMorgan Chase is reflected in the speed and variety of the studio's output. Based in the United States, Shade currently works as a software engineer at Forward Research, the lead organization behind the Arweave protocol development.
While many AI labs focus on model training, DPS Labs focuses on the periphery: the documentation systems, the context managers, and the decentralized storage layers that make AI useful. The lab is also notable for its explicit philosophical stance. Shade frames his work as building technology that empowers 'purposeful living,' a rejection of the attention-economy mechanics common in modern software. This leads to a product design language that is utilitarian and focused on task completion rather than engagement metrics.
Documentation aggregation pipeline for AI training.
AI context management with CLI, TUI, and Raycast interfaces.
DPS Labs is hiring