I Automated Myself Out of a Job (And Into the Job Finance Leaders Are Supposed to Be Doing)

As a finance person at a startup, the first time I wrote a SQL query that generated my company's income statement straight from Postgres, I felt like a superhero. That was my unique skill set — being the CFO who codes.

I have a core philosophy: finance is only as good as your accounting, which is only as good as your data. And at a startup, most data is crap. So upon joining a new company, I'd spend most of my time building data flows — clean pipelines, repeatable processes, structured infrastructure.

The problem, as anyone who works at a startup knows, is your time is limited. While I loved my data flows, they often came at the expense of what I should have been doing: high-level strategy. I was the textbook example of letting perfection be the enemy of good. If data existed, I wanted it clean and repeatable.

The thing is, AI now lets you be both.

My SQL Is Your English

For years I had this awesome data warehouse. Allocate payroll and post reversing journal entries across classes? One SQL query straight to the GL. But I was the only one who could do it. SQL looks like English — and to those who write it, it makes sense — but it's complicated to everyone else.

The tech stack I built isn't obsolete. It's more relevant than ever. It's just that now everyone can access it.

Other executives can ask questions like:

"What's our cash runway if we delay hiring by 90 days?"

And get a formatted answer with scenario comparisons before they finish their coffee. They're writing SQL queries. They just don't know it.

Everyone's a Superhero Now

That superpower I spent years building — going from raw data to strategic insight in minutes — my entire team has it now. In plain English.

Budgeting went from a quarterly fire drill to a live conversation. Stakeholders check their own numbers on their own time. Scenario planning happens on a Tuesday afternoon, not a board prep weekend.

When everyone can model, everyone thinks like an owner.

And here's the part that should make you nervous if you haven't started yet: the data is the moat. AI is only as good as what you feed it. Point it at messy data and you get garbage faster. But build clean infrastructure — structured pipelines, well-modeled data, clear hierarchies — and AI turns your entire team into analysts overnight.

Don't know where to start? Ask AI. We're in that weird meta world now. "How do I build ETL pipelines from these sources to this database?" I'll post concrete examples in the future.

Why Structured Data Still Matters

"But Geoff, I thought the cool thing about AI is it's just English?"

It is. But structured data isn't dead. Debits and credits have existed for 700 years. Conventions are great — they're a shortcut. If you want to analyze all your vendor contracts, do it once, put them in a contract object. Don't constantly rescan your PDFs every time you need to ask "which contracts expire this month."

I Stopped Looking in the Rearview Mirror

Accounting has always been backward-looking. Close the books. Report what happened. Explain the variance. Rinse, repeat.

The reason I chose finance as a profession is that we're supposed to be about the future — what happens when we buy this company, launch this product, enter this market. But most of the job ended up being about the past.

Now I have the past data flowing cleanly and repeatably, which frees me to model the future. And so can you.

I'm using live operational data to drive the company forward. Real strategic planning. Real forecasting. Capital allocation decisions backed by data the whole team can see and challenge.

The job I was always supposed to be doing — I'm finally doing it. And the infrastructure I built doesn't depend on me anymore. It doesn't depend on anyone who knows SQL. It just works, and anyone can use it.

The Wake-Up Call

If you're a CFO and not changing your workflows, you have maybe twelve months before someone who embraced this passes you. Not because they're smarter. Because they stopped being the bottleneck and started building systems that scale without them.

The Tableau guy is dead. The SQL gatekeeper is dead. Data belongs to everyone who can ask a question.

The only question is whether you're building that future or waiting for someone else to build it for you.

I'm going to start posting the specifics — real code, real MCP configs, real pipelines, working examples you can take and run with. Not theory. The stuff I'm actually running right now.

If you want to stop being the bottleneck and start being the one who built the system that made everyone dangerous — connect with me on LinkedIn.

More coming soon.

Previous: "It's Not About the AI — It's About What You Feed It"