I was wrong about AI in FP&A

In partnership with

Happy new year!!

Another January, another blown budget right out of the gates.

The past few weeks I’ve had AI on my mind in a big way…

Ever since AI hit main street with ChatGPT and everyone was buzzing about it, here’s what I was saying:

• AI has been in finance for years - just look to our data analyst friends and their use of machine learning AI

• Gen AI (like ChatGPT) has no place in FP&A - we are all about precision and numbers and AI is not built for that

• Sure AI can reorganize my calendar and answer emails, but that’s not my real bottleneck in FP&A

• Okay, it can write code but I’m not running code as a finance leader and I don’t want my team running code for something they don’t understand

But my biggest miss on all this was I was thinking too small.

I totally missed how powerful AI can be now that you can combine LLM AI (Chat GPT) with machine learning AI (statistical analysis and python) to create things like AI Automation and AI Agents.

I’ve been intentionally slow to talk about this since I’m far from an AI expert. BUT I am an expert in knowing what CFOs want out of their FP&A teams. And I’ve done a TON of research on this topic in the past month to feel directionally confident.

Am I wrong on this? Maybe.

But I’ll let you be the judge…

What are AI automation and AI agents?

Let’s jump in the deep end, then come back to practical steps later…

AI automation and AI agents are a combination of AI and process automations to accomplish specific tasks with intended outcomes:

• They gather data or info

• They execute tasks based on that data

• They make decisions

• They learn from feedback

Pretty wild, huh?

But how??

AI automations and agents are a series of software tools that are used together to solve a specific problem.

For example (outside of the finance space) a marketing AI automation or agent might have access to a website traffic research tool, a set of brand guidelines and tone of voice to write articles, access to post articles a company’s website, and access to send an email on behalf of the company. And the AI agent has a goal of increasing website traffic and ranking for a specific keyword that the target customer researches.

In this instance, we’d probably call this AI agent a marketing specialist or copywriter.

The crazy part…

This AI Agent can run 24/7 creating unlimited articles for much less than a full-time copywriter per article.

The reason I used a marketing example is 1) because I’ve used this exact AI agent to write blog posts and 2) there aren’t any that I’m aware of for finance… yet…

Will AI replace jobs in FP&A?

If you haven’t picked it up by now, I’ll say it out loud:

I believe AI automations and agents have the potential to replace a significant portion of the corporate workforce in the next 5 years.

Any job with a clear objective and set of tools can be automated or given to an AI agent.

When we look at FP&A as a job, it gets tricky.

Many of the tasks in FP&A are quite unstructured and require interpersonal coordination or decision making (which an AI agent could do, but would still need oversight).

Here’s a few examples of tasks that I’m not sure an AI agent will work (at least not without oversight):

• Translating an analysis into an updated forecast projection that is agreed upon by senior leadership

• Making a decision on what forecast drivers to leverage after getting buy-in from business partners

• Deciding how much budget to allocate or sweep to/from business units

   

Most of the roles that I believe are the safest from AI automation are the ones that require interpersonal coordination and decision making.

If you find yourself in a role that is following a very clear set of steps with a very clear objective, I would highly encourage you to take 1 of 2 paths:

1. Become someone who is trusted for interpersonal coordination (working with other business leaders) and decision making (decisions about updating the forecast, etc.)

2. Become someone who can implement AI automations and AI agents into your finance team

With that out of the way, how can we become the kinds of leaders who leverage AI in FP&A to drive better insights and efficiency?

Where exactly is the opportunity for AI in FP&A?

When I look at our function, I think there is a massive opportunity in building custom software for small to mid-sized businesses.

What is custom software?

Up to this point, you’ve had a few choices for scaling your finance team’s insights:

1. Hire more people (the most common)

2. Buy a software tool (lots of regret here)

3. Create your own automations (VBA, python, RPA)

But large companies have had a 4th option for decades: custom software.

Like any software build, it traditionally takes years and costs $100k’s to build a legit internal tool for reporting automation, advanced analysis, or better forecasting.

And large companies are willing to invest that to get returns in the $MM from efficiency and added value.

What can we build?

AI has now made it possible for non-technical people to build low-code or no-code software applications. Which brings down the cost and opens up the possibility for anyone to build software.

For example, AI can create and launch the software app for you if you simply tell it what you want to build (Replit.com) 🤯

The software you build can be as simple as automations that extract values from your ERP and drops them in a weekly KPI dashboard, as helpful as an internal web app for machine learning analytics on your sales data, or as complex as an AI Agent that you can chat with and ask it to complete finance tasks.

The cool thing about this approach is that it can be as small or big as you want - you only implement what you need.

Here’s a few examples that you can steal:

• Extract the data from Oracle (our ERP), Tableau (our BI tool), and Salesforce (our CRM) and paste it into our weekly KPI dashboard. Analyze the key findings and summarize the output in an email.

• Oversee our sales data and ensure there are no anomalies or unexpected sales trends. Take this sales information, analyze it, and send me a summary along with a forecast of the next 7 days of sales.

• When each department submits their budgets to us, scan them for material changes against the prior versions and reach out to each budget owner to gather information about the change. Once they respond, consolidate that and summarize it via Slack to the CFO.

How to get started if you want to build yourself…

I’m typically a laggard when it comes to things like AI.

But for some reason, this has me on the edge of my seat.

I’ve spent a ton of time listening to podcasts and watching videos to understand the opportunity and translate it to FP&A.

But there are 3 skills that I see you’ll need if you want to build your own AI automations and agents:

1. Automations: this includes tools like Make.com, Zapier.com, and others that help you connect software to each other through APIs

2. Machine Learning: if you want to explore the analytics component of AI, you will likely want to have a basic understanding of applied data analysis/science as well as understand how to troubleshoot python

3. Generative AI: knowing how to prompt the LLM models to get the outputs you want is key to the whole system running correctly

Then once you have a basic understanding of these, you can explore how they work together to create workflows that mimic parts of your job.

For example: the Gen AI can build the web app and write the python code that does the machine learning analysis and the output is fed back into a Gen AI via automation which gives you a plain-English output of the analysis.

In the coming months I’ll dedicate newsletters and videos to showing what’s possible with these 3 skills when combined. It won’t be polished with 100% answers, but will be more about us learning together.

I’m super excited to explore this with you!

In summary:

2025 will be the year of AI agents (my bold prediction).

And while many use cases are emerging in marketing and sales (rightfully so) I believe there is huge upside in FP&A.

If you consider yourself scrappy FP&A professional, this is a GREAT time to learn all about AI automations and AI agents to apply to FP&A.

Say hi 👋 on LinkedIn