What We Help With
Agentic coding tools that write, refactor, and debug code autonomously, plus inline autocomplete for faster development. We set up Claude Code, GitHub Copilot, Cursor, or local alternatives, and have experience integrating these into R/Posit and restricted government environments.
Make internal documents searchable with AI. Answers grounded in your data, not generic responses. Fully local deployment possible, so data never leaves your servers. Built with ChromaDB, DuckDB, and rchroma, our open-source R package.
Ask questions of databases and datasets in plain language. Create plots, filter data, generate summaries by describing what you want. Non-technical team members can explore data on their own, no coding required. Built on blockr.ai, our open-source framework.
How We Work
From first conversation to running system.
Featured Projects
Cloud vs. Local
AI Insights
Practical examples of working with LLMs in R
David Schoch, Christoph Sax /
R with RAGS: An Introduction to rchroma and ChromaDB
LLM/RAG/R
Large language models (LLMs) are developing rapidly, but they often lack real-time, specific information. Retrieval-augmented generation (RAG) addresses this by letting LLMs fetch relevant documents during text generation, instead of just using their internal—and potentially outdated— knowledge.
Christoph Sax /
Playing with AI Agents in R
LLM/R
It's local LLM time! What an adventure it has been since I first started exploring local LLMs. With the introduction of various new Llama models, we now have impressive small and large models that run seamlessly on consumer hardware.
Christoph Sax /
Playing with Llama 3.1 in R
LLM
Meta recently announced Llama 3.1, and there's a lot of excitement. I finally had some time to experiment with locally run open-source models. The small 8B model, in particular, produces surprisingly useful output, with reasonable speed.