Crafter Station is active in the middle and upper layers of the AI agent stack. They build the interfaces and infrastructure necessary for agentic workflows to function in real-world applications. This includes developing streaming UIs that can handle multi-step agent actions and building semantic search systems (like their lupa project) that provide the retrieval-augmented generation (RAG) capabilities agents need to access private data.
For builders, Crafter Station is relevant because they specialize in the practical deployment of agents, including the development of evaluation frameworks. They help teams move from a basic prompt to a production system where agent behavior is measured and reliable. Their work on inference infrastructure and data plumbing—using tools like Postgres and edge runtimes—ensures that AI agents can execute tasks with low latency across different surfaces.
Crafter Station is a software delivery studio founded in late 2023 that operates with a focus on speed and open-source contributions. Unlike traditional development agencies that prioritize headcount, the company emphasizes a "senior operators only" model. The team is distributed across Latin America, with core hubs in Lima, Peru, and Bogota, Colombia. They represent a new breed of services firms that act as an external product team for AI startups and enterprises, rather than just a staff augmentation provider. This approach allows them to ship full-stack web products and AI features with a level of polish typically reserved for internal teams at Series A or B startups.
The company is perhaps best known for its portfolio of open-source projects, which have collectively amassed over 3,200 stars on GitHub. These projects are not just side experiments but serve as the building blocks for their commercial work. One flagship project is tinte, an AI-driven theme generator for VS Code and Shadcn UI components. Another is lupa, a semantic search tool designed for codebases and structured documentation. By building in public, Crafter Station establishes technical credibility within the developer community, which in turn feeds their pipeline for product engineering and design system consulting.
Speed is the central tenet of the Crafter Station methodology. To achieve this, the team adheres to a strictly curated tech stack that includes Next.js, Bun, Vercel, Drizzle, and Postgres. This standardization allows them to move between projects without friction. In the AI space, they specialize in streaming LLM interfaces, agentic loops, and inference infrastructure. They build using the Vercel AI SDK and the latest frontier models, focusing on patterns that remain stable in production environments. Their work includes developing "evals"—frameworks for testing and measuring the performance of AI outputs—which is a critical requirement for companies moving beyond simple chat prompts.
Crafter Station is part of a growing movement of high-end technical talent in Latin America. By positioning themselves as a center of excellence for AI engineering, they provide a bridge for North American and European firms to access specialized talent in similar timezones. The company is deeply involved in the regional ecosystem, hosting events, hackathons, and "Code Brews" to connect builders. This community-first approach has turned the studio into a hub for technical talent in Peru and beyond. For clients, the value proposition is simple: they offer the velocity of a startup with the technical rigor of a specialized AI lab, bypassing the overhead and "nonsense" associated with larger consulting firms.
AI theme generator for VS Code and design systems.