Skip to content
Work

Real builds, shown as they actually stand.

One project has a public repository. The rest are closed-source builds we are happy to walk you through on a call. Status is marked on each.

Agentic AIIn development

Sales intelligence briefing agent

Give it a company name and it researches the company the way an analyst would: scraping public sources, cross-checking claims against the page they came from, and streaming its reasoning as it goes instead of returning a static report. Nothing reaches the output without a citation it can point back to.

  • LangGraph
  • Pydantic v2
  • FastAPI + SSE
Agentic AIClosed-source build

SketchUp MCP

An MCP server over SketchUp's Ruby API: it lets an AI agent read, create, and edit a 3D model in an open SketchUp session through the Model Context Protocol, instead of a human driving every click by hand.

  • Model Context Protocol
  • SketchUp Ruby API
  • 3D Automation
IntelligenceClosed-source build

Wildfire risk prediction pipeline

Pulls Sentinel-2 satellite imagery from Microsoft's Planetary Computer catalog and turns it into a wildfire risk read for a given region, built to flag the areas worth watching before a fire starts, not after.

  • Sentinel-2
  • Planetary Computer
  • Geospatial ML

Detailed case studies are being written.

Full write-ups and recorded demos for each project are in progress. In the meantime, the fastest path to detail is a short call — we can walk you through any of these directly.

Start a conversation