Building at Tightly
The craft of building an AI-powered supply chain product: how the forecasting engine works and should be evaluated, what a principled 12-month roadmap looks like, and the PM methods that actually fit a team of under 30 people building software where mistakes have real financial consequences.
Tesuji is the clever local move that solves a tactical problem — the unexpected sequence that captures stones, creates territory, or turns a losing position around. In product development, tesuji is the specific technical or methodological choice that makes the product distinctively good: not just 'we use AI' but the precise evaluation method or design pattern that makes the forecast 12% more accurate than the next tool.
Demand forecasting is a fundamentally different AI problem than chatbots or classifiers. Here's what the PM needs to understand about how it works, how it fails, and how to evaluate whether the model is actually getting better.
What a principled 12-month Tightly product roadmap might look like — organized into strategic themes with the retention logic and customer segment rationale behind each one.
Supply chain PM is different from consumer PM and from growth SaaS PM. The decision loop is slow, the user base is small and expert, and mistakes have real financial consequences. Here's the method set that works.