What we make
Software we build, run and use ourselves. Both are early, and both are open to a small number of teams.
Demand planning that shows its working.
Forecasts by item and location, with the drivers and the uncertainty in plain view — so planners can adjust with confidence instead of overriding blind. Demand first; inventory, supply and S&OP next, on the way to end-to-end supply chain planning.
A Python desk for long-running AI agents.
A persistent, controlled Python runtime agents reach over MCP. Packages, files and processes stay put between sessions, for days or weeks — no servers to babysit.
agent → python.exec("import pandas as pd") agent → python.exec("df = pd.read_parquet('orders.parquet')") # next day, same desk agent → python.exec("df.shape")Visit pysmith.com ↗