Early in my career I used to dream up these wild ideas and bring them to my boss.

“We can’t get the resources to build that” was always his answer.

To be fair, they were definitely nice-to-have things compared to what our developer team was working on to actually impact the business.

One of those random ideas was an interactive phone/web app that our senior leaders could use to get a quick glance at results - specifically styled like the stock market app on an iphone.

Rather than senior leaders waking up to an excel or pdf of results, or clicking through PowerBI on their phone, it would be way better to have an app that was user friendly.

The problem was the value to effort ratio - not enough value compared to the effort it would take.

Fast forward to 2026 - while I was reminiscing on that experience and writing the words above, Grok was on my other screen building the exact app I dreamed of years ago.

You can’t tell from here, but it’s fully interactive (fake data).

But with another 15 minutes and a few more prompts, I could make this into a fully-functioning app for a business.

And now if I can dream it I can do it, I’ve been having fun 10x’ing my analytics with Claude Code and thought I’d share exactly what that’s looked like for me as VP of FP&A…

I used to sit in meetings all day, intake work, push it to the right person, then see a first draft a few days later, iterate for a week, then send out a first draft.

Now I sit in meetings all day, intake work, spin up a new Claude Code session, check back in 5 minutes, and send out a first draft in hours.

It’s wild how much work I can just handle myself. And how much closer to the data/business I feel in 2026 using AI than I did just a year ago not using AI. For those willing to learn some new tricks, it’s totally worth it.

And having a full team of analysts (including myself) using AI like this, it truly changes how much we can output in a week. The 10x claim is no joke.

And there is 1 single tool that was the unlock for us…

I’m not a software engineer, so when I was first introduced to Claude Code I was a bit overwhelmed.

But with the help of our IT/Data function, I got set up with what I thought would be overkill or (at worst) a waste of time.

I couldn’t be more wrong.
Claude Code is my key to the 10x productivity.

But it’s not just Claude Code itself, it’s the 4 layers that it represents which make it powerful:

If your head is spinning at this, don’t worry - mine was too. But over the next few weeks of newsletters I’ll be deep-diving these and my first workshop in October will cover this in-depth. Look out for an email invitation in the next couple weeks!

To be clear, data infrastructure/strategy is the job of data analytics to govern, design, and implement this. However, a great FP&A team should be sufficient in leveraging this working model to generate their work.

I want to keep this super practical - let’s jump into 5 things I’ve been doing with Claude Code to 10x my analysis output.

Remember, the goal of FP&A is impact - specifically impacting the business results. Historically we’ve had an issue with speed-to-output but these 5 proof-points show you how to blow up any speed-to-output issues you might be facing:

Don’t you love when a business leader throws out a hypothesis in the form of “I wonder if x is driving results for the month”?

You know it’s not true, but you now have the burden of proof to respond. And traditionally this would have taken an analyst a full day to run down the answer.

With a solid Claude Code setup, it can be as easy as spinning up a new session and having it pull the data directly from the data warehouse.

Will you get a perfect answer? No.

But will you have enough to clearly show that Labor Day is (again) not the driver of sales missing their goal this month? Yes. And it happened in minutes without distracting my team.

Back in the day I used to block out hours to study results - I would encourage my analysts to do this too.

I’d focus on revenue and the sales funnel to see if anything is shifting in the past week.

I recently hammered out a massive prompt in Claude Code explaining exactly what I was doing, which reports I was looking at, how I evaluated if there was an issue, etc.

It took some back-and-forth, but I eventually built a /weekly-funnel-review skill that I trust as my weekly review. This literally saves me hours per week.

I simply prompt “run /weekly-funnel-review for the week ending 9/25/26” and it’ll pull the 7 days of data and compare it against a 13 week baseline of sales/revenue by every cut that I want. Then the output is an html I can send to business leaders.

And it was the first to catch an upstream data issue impacting how sales get categorized to each channel.

I talked a bit about this last week, but I’ve built multiple reports that the business has needed in hours (not days or weeks).

One particular success was using our call data into our service center - we used the report to trend every standardized call metric directly from our data warehouse with drill downs by coaching team (Dialpad’s terminology).

I pushed it to a microsite that could be shared and accessed and refreshed for the business leaders.

The beauty of this is that I (as the VP) could then iterate with direct feedback - no game of telephone with analysts or others not in the room.

This one is clutch especially for those on your team onboarding or shuffling responsibilities…

But you can design and build a skill (/analyze-excel) that studies and breaks down an excel file/report/model, telling you all the sources of data, connections, building documentation, and then providing recommendations on how to improve it.

The real benefit here is running this skill as a first step to create the context for then improving/changing the report.

Once example was actually that same call metric report I mentioned before. I ran the analyze-excel skill on a previously build excel report in the pre-AI era, then told Claude Code “rebuild this report but use SQL to write the script to pull data out of the data warehouse, save the SQL into the report, along with instructions so you can update it again”.

We took a report that was built over dozens of hours last year and recreated it with a SQL to our data warehouse in an hour - and it can now be updated with a single prompt… all because we had a skill to tear the original report down to the studs.

If you’ve made it this far and this is new to you, you’re feeling the weight.

When you do something new like upend over a decade of how you work, it’s heavy.

But you have a secret weapon - you have Claude. With infinite access to how this all can work.

Not sure how to create a Semantic Layer… Ask Claude.
Not sure how to structure your files to work with AI... Ask Claude.
Not sure if you have conflicting metrics in various reports... Ask Claude.
Not sure if you can turn something you just did into a skill… Ask Claude!

Maybe the biggest unlock I found was that where I used to hit a wall and I’d think “oh well, maybe that’s where this analysis ends” has turned into “I’ll see if Claude can figure it out”.

Say hi 👋 on LinkedIn