AppliedAI

Overview
Whether you’re predicting grocery store shelves or hospital inventories, forecasting demand for thousands of individual items presents a unique set of challenges. In this talk, we’ll explore the world of hierarchical demand forecasting through real-world examples spanning everything from cream cheese to blood thinners. Along the way, we’ll cover the practical challenges of building forecasts at scale, the trade-offs involved, and the techniques that make these systems accurate, robust, and production-ready.

Overview
Every Process is a Scar: Rethinking the SDLC for an Agentic Future

Every process your team follows is a scar: an inherited response to a past wound. We build process to compensate for how people work. We forget, we tire, and we leave, taking hard-won knowledge with us. But when a new kind of actor, an agentic AI, begins working inside the software development lifecycle, the right question is not, 'How do we carry these processes forward?' It is, 'Why does each process exist, and does that reason still hold?'

This session uses an engineering constitution, a system in which every rule must carry a reason, as the worked example. It examines four areas that teams should reconsider as they build an agentic SDLC:

• People and the scars behind the process. Why teams create process in the first place, what human limitations it is designed to address, and which assumptions change when AI agents join the delivery model.
• Process from first principles. How to interrogate each practice on its own merits instead of inheriting it by habit or relying on “we have always done it this way.”
• Tooling with reasons built in. How an engineering constitution can connect each rule to its purpose, so the rule can be re-examined when the context changes.
• Scaling what still earns its place. What it means to grow a team while retaining only the processes and guardrails that continue to justify themselves.

This is not an argument for fewer guardrails or for burning everything down. A process may absolutely deserve to stay, but it should re-earn its place through first principles of reasoning rather than tradition. Attendees will hear real experiences, real scars, and honest reflection on what was kept, what was questioned, and what building an agentic SDLC forced the team to rethink.

What the audience walks away with

• A practical framework for questioning inherited SDLC processes as agentic AI changes how work gets done
• A clearer way to distinguish guardrails that still serve a purpose from process that persists mainly through habit
• An engineering constitution concept that links every rule to a reason and makes future re-evaluation possible
• Questions attendees can take back to their own teams to challenge process from first principles without adopting someone else’s model wholesale

Format
35 to 45 minutes, single speaker, no product pitch. Q&A as time allows.

Minnesota's biggest AI event of the year is back. One day, multiple stages, every format you can think of — keynotes, panels, workshops — covering how AI is actually being used across industries.