Workflow Wednesday: ⚙️ Cut Artificial Complexity Like Michelangelo

Art by @basilonmypizza: https://lnkd.in/eF8FkWzN - https://basilhefti.ch/

Inherent complexity is the difficulty in the business problem; artificial complexity is the extra friction we add through design and org choices. Think Michelangelo: the figure is “already in the marble,” and the work is to remove what is not essential.

When integrating AI into real workflows, we address business problems. Inherent complexity is everywhere: global suppliers and shifting regulations; customers moving across web, chat, store, and app; fraud defenses that blend live signals with deep history.

And we add artificial complexity. Departmental data silos. A mesh of microservices with no end-to-end owner. Bolting AI onto “how we have always done things.” Incentives compound it: local KPIs, fragmented compliance, duplicated budgets.

The outcome? Fragile systems, overruns, missed value.

How to fix this? Separate artificial from inherent complexity. Count handoffs, services, data stores, queues, and repos; list owners for each interface; mark anything that exists “because our org chart says so.” If the design rationale is not clear, remove it.

Keep It Smooth & Simple (KISS): prefer one storage bucket unless a documented need exists; consolidate data in one catalog rather than many (e.g. on Databricks); govern access with role based and attribute based access control instead of container-level segmentation; keep a monorepo unless a split is clearly justified; and so on.

Complexity will always be there. Artificial complexity is self-inflicted.

Be Michelangelo: cut away what is not needed.

👉 How do you keep artificial complexity in check when deploying AI?

• Happy Birthday Markus Vitali 🎂

• Art by @basilonmypizza: https://lnkd.in/eF8FkWzN https://basilhefti.ch/