Why I'm Glad I Went All-In on AI in 2025
In early 2025, I made a bet. I stopped thinking of myself as a software engineer and started thinking of myself as someone who manages AI to build and maintain technology. A year later, I'm convinced this was the right call. The writing was on the wall. When I first started experimenting with Claude and Cline in January 2025, I was paying 5 cents per question to have AI analyze codebases and generate solutions. By mid-year, I was building production systems where AI wasn't just helping me code—it was the product.
The Shift
Traditional software engineering—the kind where you spend hours writing CRUD operations, debugging edge cases, and maintaining legacy systems—is entering its twilight. The new skill isn't writing code. It's knowing how to orchestrate AI systems to do it for you, and more importantly, knowing when they're doing it wrong.
At my company, we went from zero AI products to four production systems in one year: AutoDocs Engineer for multi-modal troubleshooting, an autonomous Service Agent processing customer requests, a Sales Enhancement recommendation system, and automated quote surveys. Each one was built faster than traditional development would have allowed.
What "All-In" Actually Means
Going all-in wasn't just about using Copilot for autocomplete. It meant restructuring how I think about problems. In my 2025 roadmap, I identified three areas: Product, Workforce, and Automation. The common thread? Every department should be able to automate their processes with AI—not just IT.
We added computer vision to our troubleshooting system. We built MCP integrations connecting calendars, email, and ERP systems to AI agents. We even started letting AI build the tools it would use itself—proving out the concept with DraftSite and SolidWorks automation.
The Future Isn't Code—It's Context
The engineers who thrive in the next decade won't be the ones who can write the cleanest functions. They'll be the ones who can provide the best context to AI systems, validate their outputs, and architect the workflows that tie everything together.
Software engineering as we knew it? It's not dead yet. But it's on borrowed time. I'm glad I didn't wait to find out.