About a year ago, I was pitched Claude at a Taylor White CPE event as a potential replacement for Blackline. At the time, my company had Blackline, but we just weren't using it right. We would upload GL detail and call it an account reconciliation. Task management sat unused. It was just another list on top of the Excel checklists everyone already kept for month-end close.

I went to our CFO, got approval for four Claude licenses, and stood up a pilot on my team. We held weekly sessions to share what we were learning. It felt promising.

It was also harder than I expected. It stayed hard for months.

Learning to prompt well is a skill, not an instinct. People would try Claude once, not get the answer they wanted, and quietly go back to the old way of doing things. It's hard to keep someone motivated through trial and error when they're already underwater with their actual job. I started to feel like a broken record: Did you try Claude? Did you ask it? What did it say? And the hardest part? I was struggling right along with them.

I tried everything I could think of. I tried reconciling data sets, but with less than perfect data, I hit a wall. I tried using it for deeper analysis, but Claude didn't understand our data and gave inconsistent and vague responses.

So, I scaled back to what worked: writing emails, performance reviews, first drafts. Claude saved me real time there, no question. However, I was still drowning in manual processes, while still needing to close the books faster with better variance explanation.  I was still not touching the "transformative AI" I kept hearing about. I blocked an hour a day to keep experimenting. Some weeks I honored that block…once.

Then one night, my husband mentioned how much he loved Claude. He could just type a slash command, and it would create a Jira ticket for him automatically.

I had no idea what he was talking about.

The next day, I went to work and asked Claude to explain it. That single question cracked open a part of Claude I didn't know existed, its coding capabilities. I know that probably sounds obvious to anyone already deep in this world, as it's one of Claude's headline features. I was an accountant who had been told AI was "game-changing" with zero roadmap for what that meant. Nobody around me was an expert. Nobody was showing me the edges of what this tool could do. I had to stumble into it myself.

My next prompt was simple: give me 3-5 use cases of what I could automate as an accountant using Claude’s coding capabilities.

That question unlocked everything.

First, I had Claude write a script to read our close checklists and generate an executive summary automatically, cutting down the long, daily standup we used to run just to report status during close.

Then came payroll. We had more than fifteen payroll entities, most paid bi-weekly, each requiring its own journal entry creation. Every cycle, we had to pull data from ADP, remap department codes to match our ERP, reformat everything, and only then upload it, all done manually by our offshore team, at roughly an hour per entity, per upload. We had previously tried to buy software to automate this. But with how differently each entity formatted its data, off-the-shelf tools couldn't handle it. Claude helped me write a program that held the full ADP-to-NetSuite mapping logic and generated the upload file directly, turning that hour-per-upload process into seconds, with an audit trail I could actually explain because I understood every step of the logic myself.

That was the real unlock. We were a company built through acquisition, every entity had its own quirks, and nothing was standardized. I kept hearing that Claude was made for analysis, but we could barely get to analysis, because manual processes were stretching our close to ten days. The Jira comment sent me down a rabbit hole that ended up somewhere much more useful: Claude wasn't just a writing assistant or an analysis tool. It could automate the manual work sitting between us and clean data, and clean data was the thing standing between us and the analysis we were actually hired to do.

If you're an accountant experimenting with Claude and only getting value out of writing tasks, you haven't hit the ceiling. You haven't found the front door yet. Ask it directly what it can automate for your role. Ask it to explain a capability you've only heard about secondhand. The gap most finance teams face isn't that AI can't help with the manual grind. It's that nobody's told them to ask.

If your team has underused AI tools or is struggling to understand what is possible, we have been there, and that is exactly where we can help.  Our AI leveraged Advisory Service team is comprised of seasoned accounting & finance professionals with a passion for transforming the way you work.  We’re not “just” IT or sales professionals.  We’ve sat in your seat. We are process & controls minded. We are senior experience that executes.  Reach out for a complimentary consultation & advice session today.  info@taylorwhite.com