🤖 Run, the bots are coming

Breaking down the future finance function, one department at a time

Sponsor header
Runway

Financial planning should take minutes, not days

With Runway, it does. Whether you’re adjusting hiring timelines, modeling expansion plans, or testing "what if" scenarios, Runway delivers real-time clarity in seconds. Just drag and drop plans to see instant changes in metrics like cash flow, margins, and runway.

This isn’t just talk. Superhuman tracks growth with real-time insights using Runway. AngelList relies on it for smarter, faster modeling. And 818 Tequila keeps their teams aligned without the grind of spreadsheets.

Runway transforms finance teams into strategic powerhouses, bringing your whole org onto the same page.

Working Capital header

We’re still so early

“That’s insanely cool. Please show me that again.”

It was my first time seeing an RPA (Robotic Process Automation) tool firsthand.

My head of AR (Carole) showed me how they had been testing ‘task bots’ to allocate cash received to individual customer sub-ledger accounts.

We had a team of twenty people doing this manually every day. It was time-consuming, costly, and occasionally prone to error.

She showed me again how the task bot would look up the cash receipt, find a matching remittance in the inbox, break down the payment into invoices, and then allocate those payments against invoices in our ERP.

“What happens when it can’t tell which payment is from which customer?”

Carole smiled: “That’s the coolest bit. You go in and find it manually, and show it back to the bot. But once you’ve done that, it will learn from its mistake and code it correctly next time. It’s called machine learning.”

“COOL! How much of your team’s cash allocation workload can these things handle?”

“This is just a small trial on a freebie from the supplier, so not much. But theoretically, I reckon it could do 90% of the work. Just need you to approve the CAPEX spend… they aren’t cheap.”

“Haha, I figured there was a catch. Well, I’m on board. Can we use these for bank reconciliation, accounts payable, and expense processing, too?”

Carole: “That’s not my world, as you know, but I don’t see why not. They are really good at anything that is purely processing to a given rule set. There’s just one drawback…”

“What’s that?”

Carole paused for dramatic effect, then chuckled. “One of the bots just put a request for some PTO at Easter.”

Over the next few years, we reduced the cost of our transaction processing by 50% by rolling out bots across the transactional accounting functions. But the dollar saving wasn’t the biggest benefit. It was the speed of work and elimination of manual errors.

And this was ten years ago, so is just a fraction of what will be possible with the new generation of AI tools.

Transform finance teams into strategic powerhouses, bringing your whole org onto the same page.

Deep Dive header

Run, the bots are coming

Welcome back to this Playbooks series on The Future CFO.

Last week, we debunked some myths about the future of the CFO role and mapped out the real forces driving change in finance teams.

I believe we’re in the eerie calm before the tsunami hits. The tide has pulled back, and an AI wave is about to come crashing through, disrupting everything in its path. And by ‘everything’ I mean accounting and finance. We are first in line to have our ‘landscape transformed.’

This week, we’re diving into how that wave will reshape the teams beneath the CFO. What gets automated and what stays human.

Or at least, my best guess of how it will all play out …

Before we start, a few disclaimers:

  • I don’t have a crystal ball. If you’re reading this in 2030, you’ll probably cringe at how wrong I was.

  • I’m not an AI / tech nerd. But I am using LLMs (ChatGPT, Claude, etc.) every day - mostly replacing Google (and more). So, I’ve got a good hands-on feel for what AI can do and the pace at which it’s improving. Spoiler: it’s happening at rocket speed.

  • I am a finance nerd. I know what’s automatable, what isn’t, and where the real friction points in adoption could lie.

And what time horizon am I assuming? This is guesswork, but let’s go for ten years. No one knows how long this will take. AI labs talk about Artificial Super Intelligence in 5-7 years.

But finance teams aren’t AI labs. Finance moves slower. Not because the tech isn’t ready, but because trust, adoption, and regulation lag behind.

I believe ten years from now, the most aggressive automation will have happened, but there will still be holdouts. The last mile (trusting AI in high-stakes decisions) takes time.

What AI won’t replace

AI will eventually automate almost everything in the accounting and finance functions. But some things resist automation. Not because the tech can’t do it (or at least try), but because people won’t let it.

  • High-stakes judgment calls: AI will support modeling the scenarios, but someone still has to sign off. As a CFO whose job and reputation are on the line, I’d be hard-pressed to hand over high stakes decision agency to AI.

  • Accountability & ownership: You can’t fire (or shame) a bot… yet

  • Relationships: The board doesn’t want a chatbot pitching the next acquisition. Investors will want to talk to management, not a computer.

That’s a short list. Everything else? It’s on the chopping block.

Y Combinator is already calling for founders specifically to build AI startups that disrupt back-office functions - finance, legal, and HR. The tech is coming, fast.

Many finance pros are (rightly) concerned about privacy and compliance. But this take ignores the exponential nature of breakout technology development. We need to think about the LLMs we have seen so far (ChatGPT, Perplexity, Claude, etc) like the internet in 1993 or iPhone in 2008.

They are a platform.

A platform on which people smarter than you and I will build things we can’t imagine yet.

So the real question isn’t when will the tech be ready - but how fast will it be adopted.

Bye-bye big bang

When I first rolled out RPA bots ten years ago, it was a big program. We had a big capex project, Project Management Office, a “big bang” rollout strategy, and months of planning.

This is different.

The wave of AI tools won’t need a grand rollout. These tools will interface seamlessly with humans and systems, quietly taking over tasks—faster, cheaper, and more accurately.

And then they’ll do a little more. And a little more.

But what makes this wave truly different is that AI adoption itself will be automated.

Unlike RPA, which requires humans to configure workflows, AI can self-learn, self-improve, and self-integrate into systems.

And remember, this stuff is the worst it will ever be. Tomorrow, it will be better. The day after, better still.

One day we’ll all wake up and realize the finance function isn’t what it used to be.

Now let’s see what this could look like across each part of the finance function:

Secret CFO Predictions for The Future of Finance

Financial Controlling (Accounting & Reporting)

In my day we used to have to call our customers to chase payments, can you imagine?”

Accounts Receivables and Accounts Payable functions will be first in the firing line:

  • Invoice processing? Automated.

  • Cash allocation? Automated.

  • Purchase invoice matching? Automated.

  • Credit control? AI call bots will chase late payers (and probably do it more relentlessly than your current team)

Sure, there’s some human interaction necessary here. But let’s be real, most of this is routine processing.

Yep, that means before long robots will be calling your own AP robots to argue about overdue invoices. Can’t wait…

And who knows… in the long run maybe we end up with single irrefutable ledgers for business transactions on a blockchain ledger. AI will be an eventual enabler for this, I believe… *boomers’ heads explode*.

Month-end close is also ripe for disruption.

Why? Because:

  • Journal processing follows predictable patterns

  • Balance sheet reconciliations are rules-based

  • Anomalies can be auto-flagged and escalated

AI loves anything that is highly technical but ultimately manual. Think: those complex multi-entity consolidation spreadsheets.

Tell me why any of this will require a human in 10 years. Maybe even five?

Yes, there will still be edge cases. The one big bad debt decision or the tricky judgment call on revenue recognition. But those already escalate to a senior controller today. AI just removes the 99% of routine cases that don’t need human input.

And all this will strip days out of your close. Other than leaving your books open for a few days to capture any purchase invoices in transit, why would you need to wait to close?

AI will learn to predict what invoices are going to land, tighten controls upstream, and be better at booking accrued expenses than your teams ever could be.

In that world, why wouldn’t your balance sheet be ready for review at midnight on the 31st of the month?

And what about reporting?

Today’s AI apparently could already generate 95% of an S1 in ten minutes.

Why can’t it do the same with a 10-K?

It will.

Today’s AI tools are not good at showing their work, and without an audit trail, we can’t trust it for high-stakes reporting. But this will change, that is just functionality that hasn’t been built yet. Not a capability problem.

Internal reporting is on the table too. Why mess around with formatting in Excel, or even a specialist reporting tool?

Your AI analyst can produce your numbers in a pre-populated template with a chat language interface. And probably produce a better commentary than your current team can write. The current LLMs are already very good at this.

Setting accounting policies will likely be owned by humans, for a while at least. These are the guardrails within which the AI will be asked to operate. And AI setting its own guardrails is a bit too meta for me to comprehend right now.

Audit firms are already asking management teams to sign representations that AI was not used to prepare financial statements.

But in the long run? I wouldn’t be surprised if we end up with a single trusted AI agent that prepares accounts. One that audit firms have signed off on.

P.S. If you’re building this, I want to invest.

Auditors would then exception-report any areas where management has overridden the AI.

Let’s be clear: the accounting function won’t disappear. But most of the people in it will.

95% of today’s work in accounting and reporting teams will not need a human.

What remains? The 5% of work that can’t be automated. And that work becomes more valuable than ever.

Controllers won’t just be number-checkers, they’ll be overseeing vast AI-driven finance ecosystems.

The controller skill set will become rarer, and much more highly leveraged.

The best controllers will be highly tech-savvy, deeply strategic, and in more demand than ever. They will be the most valuable asset in the accounting and finance function (more so than they are today.)

And as I have said until I’m blue in the face a controlling skill set will always be essential for CFOs.

Investors, auditors, and regulators will always need a human to hold accountable. AI might prepare the numbers, but when something goes wrong? They’ll still want someone to sue.

If you are a controller and not building these skills in AI & automation, you are a) missing out on a career-defining opportunity and b) putting yourself at risk of death by robot.

Tax

In-house tax teams? Compliance-only tax teams have no future. Same for basic tax structuring/planning.

But complex tax structuring? That’s still a game of legal nuance, aggressive interpretation, and careful risk balancing. AI can help, but I don’t see CFOs handing that over to the bots just yet. My bet:

  • Tax compliance work disappears. AI eats it and the occasional escalation gets swallowed up by the controller and their team

  • Same for basic tax structuring.

  • But complex tax structuring will move further to Big 4 firms & law firms

  • The biggest companies will still keep in-house tax experts, but fewer of them will be able to justify the investment

Tax strategy and structuring will always have a human element, but only at the top end where relationships really matter. Most of this kind of work today is outsourced to accounting firms for all but the biggest companies in the world.

Internal / Corporate Audit

Audit is interesting and the most difficult to predict, I think. At face value, routine balance sheet audits, control work, and S404 testing is highly automatable. Fertile robot territory. See above.

But AI is good for fraudsters too. And the risks are growing. There are already countless stories of accounting teams being fooled into making payments by deepfake videos of their CFO.

Friendly reminder: those are only going to get better.

I can see the audit team getting deeper into fraud prevention and investigations. Internal auditors of the future will have to be WAY more tech-savvy than the current cohorts.

FP&A

Here’s where things start getting juicy.

Current FP&A use cases for LLMs are limited by their inability to show the audit trail behind their reasoning. But much like the privacy and compliance problem, this is temporary. Once the nerds turn their attention to solving it, we’ll forget it was ever an issue.

I’m going to crudely dissect FP&A into three sub-functions:

  • Performance analysis & reporting

  • Budgeting & financial planning

  • Strategic decision support

Let’s start with the easy one.

Performance Analysis & Reporting: Swallowed by the Machine

Anything backward-looking (performance reporting, variance analysis, financial dashboards, etc.) will get swallowed by the automation stack under the controlling team.

I expect this will evolve into full self-service reporting powered by natural language AI.

The need for finance analysts to produce ad hoc information requests for the business will be gone soon. If it’s past-looking, rule-based, and repeatable, then it’s bot-fodder.

If the VP of sales wants to know what their bottom ten gross margin customers were for last month. They’ll just ask their personal reporting bot built to give them whatever they need.

Budgeting & Financial Planning: AI is Taking Over the Low-Stakes Stuff

There’s a bit more nuance here.

  • A new generation of AI-powered FP&A platforms will devour budget models, rolling forecasts, and heavy spreadsheet workflows: Speed and clarity will matter more than manual accuracy

  • AI-driven forecasts will be fast, iterative, and more real-time than they are today (I’m not sure this is a good thing, but I think it’ll happen)

  • Does the business really need a human partner for low-stakes, operational forecasting? Probably not.

If a plant director wants to know whether they should run an overtime shift on Saturday, they won’t ask finance. The Operations AI Bot will crunch labor costs, demand, and inventory levels, and give them an answer in seconds.

But what if they want to redesign shift patterns altogether? Or propose a multi-million-dollar CAPEX program to expand the plant?

Now we’re in higher-stakes territory…

Strategic Decision Support: The Last Humans Standing

AI can model anything, but it doesn’t have instincts.

The best strategic decisions? They happen in rooms full of smart people, debating messy trade-offs:

  • New market entry

  • Restructuring business units

  • M&A evaluations

  • Shifts in long-term strategy

No one wants to make those calls alone. No one wants AI to make them, either. The AI will bring the numbers. The CFO (and team) will bring the finance judgment.

High-stakes decisions are best tackled with slow thinking and a roomful of smart people. I don’t think that changes. And the closer you get to the C-Suite, the more true this becomes.

What will the FP&A function look like?

I expect some current FP&A tasks will shift to the controlling team because that’s where the automation and AI expertise will live.

But the more strategic end of FP&A? That will splinter off and probably get rebranded. Something like Strategic Finance.

It’ll be deeply embedded in the business, working more closely with the business. Less spreadsheet jockeying, more real business impact. Everything FP&A should have been all along.

And this will coordinated through a single universal ‘model’ for the business. Rather than the model being a spreadsheet that belongs to finance, it’ll be an omnipresent model that belongs to the business. It will represent the collective understanding of how the businesses’ finances work.

And ‘strategic finance’ will be the in-house experts and consultants for that model. No more silos.

Treasury & Investor Relations

Basic treasury operations will get automated and likely consumed by the controlling function.

Payment processing, cash sweeps, and basic hedge management is all manual stuff today with a very low dependency on judgment and relationships.

More sophisticated transactions will need deeper expertise and relationships, especially as we come to managing banks and investors. And their questions will get better, as they themselves are armed with AI,

So treasurers will need to get closer to both their investors and the business. Real subject matter specialists. And with the operations automated, they’ll have the room to do it.

Corporate Development

M&A is about relationships, not about modeling.

I can see two directions here: 

  1. In-house corporate development pros build investment banking quality industry networks and outsource their modeling legwork to AI and transaction execution to advisors, or…

  2. The whole function gets outsourced, and the CFO just gets their favorite investment banker on a retainer to manage future M&A for them

Time will tell. I expect the latter is more likely, and any in-house work needed just gets swallowed into ‘Strategic Finance'.’

Net net

You can see the extent of the disruption that is upon us. If it’s not relationship-based or doesn’t require high technical judgment, it is vulnerable.

But the work that is left behind will become even more special. More valuable. Most of that work will be consolidated into much leaner, but more specialist, ‘controlling’ and ‘strategic finance’ teams.

And where does that leave the CFO?

Well… that’s what we’ll talk about next week!

Next week we’ll get more into exactly how CFOs need to adapt.

PS - My views on this topic are evolving rapidly, so I will revisit it from time to time through future posts.

Footnotes header

:::::::::::::::::::::::::::::::::::::::::::::::::::
:: Thank you to our sponsor ::
:: RUNWAY ::
:::::::::::::::::::::::::::::::::::::::::::::::::::

What did you think of this week’s edition?

Login or Subscribe to participate in polls.

If you enjoyed today’s content, don’t forget to subscribe.

Advertise with SCFO

Reply

or to participate.