I spent the first 15 years of my career on the finance and accounting side of small, scrappy manufacturing firms — think $100M in annual revenue. These are the kinds of places where you get to (or have to) wear many hats, and learn to work with what you've got. Systems are outdated, IT support is thin, and if you want to automate a process or alleviate some pain, you'll need to figure it out yourself. Spreadsheets, macros, and ungoverned database operations are the norm.
There are so many ideas that would make life and business better, but you're largely on your own, and building solutions gets relegated to weekend hobby work. You can't step away from the machine to build a new machine.
From there, I jumped to the other end of the spectrum, joining one of the world's largest privately held companies: Koch, Inc. Koch operates by principle-based management, encourages entrepreneurial thinking, and drills a transformation mindset into each of its 130,000 employees. And by every measure, they are extraordinarily successful. However, with business and processes at that scale, transformation comes with its own challenges. There are delivery teams, business teams, capability teams, and infrastructure teams, each with their own roadmaps articulated in months and years.
Solutions often felt outdated by the time they were delivered. Even sitting on a transformation team, with ideas mapped out and tools in hand, the movement could feel like pond water.
So after I met the CoPlane team and got to know what they were after, I decided to make the jump and join them as a forward-deployed engineer: embedded directly with the customer's team to build, ship, and operate production-grade AI workflows in weeks, not quarters.
The proof showed up in my second week. I replicated an enterprise invoice processing solution I'd helped build previously — a project that had taken six months and was celebrated for its speed. The rebuild took a week.
A more recent example: a customer was manually managing a daily and end-of-month cycle count process across 60 warehouse sites. Every day, someone was spending hours tracking down each site to confirm they'd submitted their counts, chasing discrepancies, consolidating results, and reporting up to finance. I spent two days building an AI application that handles all of it: proactively prompting sites to submit; reconciling variances against production software, vendor shipment data, and ERP records; and routing exceptions and approvals by rule. The app runs all day, every day, and procurement can go back to procuring.