The following column is by Siqi Chen, CEO of Runway Financial. The opinions belong to the author.
In the first half of 2025, businesses eliminated thousands of roles.
Most of the statements sounded the same: streamlining operations, reshaping cost structures, leaning towards AI. It was difficult to discuss logic. The numbers have been checked out. And from afar, many of the roles that have disappeared may seem redundant.
Funding was one of them.
But the thing about finance is that it's always looks It can be automated from outside. Assume that you will see data, models, and repeatable workflows and that this part of your business can primarily be done.
That's a mistake.
What is funding?
Finance looks mechanical: reporting, budgeting, final forecasting, forecasting.
But the actual work happens below.
This is the only function trained to simulate whole work. This is one place where product roadmap, sales goals, employment plans, customer behavior and market volatility all come together.
Sometimes the job is to check if the numbers are summed. More often, it's about answering strategic questions. What happens if I pull this lever instead of this lever? If I'm delayed, what will break? What is it worth doing?
Spreadsheets are not values. The reasoning is.
So you can pass it to a machine that is too fast because it costs you to see it right away.
Automation Paradox
Some of the most obvious wins in finance are the least important. Automate invoice matching, forecasting, and pipeline rollups. You can also generate board decks in AI. And it all saves time.
However, AI does not answer questions like:
- Can I take this risk?
- What will you lose by waiting for another quarter?
- How sensitive is this result to the input behind it?
AI can help you raise funds. You can point out important variables. You can also suggest the next step.
But I don't know that where What you are about to go, or you are willing to do to get there.
That's experience and judgment. And they are still human.
“One day, when someone asks why Customer acquisition costs are rising rapidly. No one knows. Because I had to explain it for months. ”
The lines cross
It starts small.
First, automate the end of the month. Then predicted. Then a scenario plan. Then, decision support.
Each saves time. Things get faster and everyone seems to be at ease.
But then people stop touching the model. They will stop revisiting the assumption. Edge cases stop logging and slowly fade intuition.
One day, when someone asks why Customer acquisition costs are rising rapidly. There's nothing no one knows anymore. Because I had to explain it for months.
A familiar pattern
When AI fails financially, it rarely appears to be an AI obstacle. However, patterns are recognizable.
- GPT ‑ 4 Turbo failed 81% of FinanceBench financial questions.
- Knight Capital lost $440 million in 45 minutes due to a trading script that no one has reviewed.
- The interactive broker lost $48 million due to misunderstandings in NYSE data.
That's what happens when humans are too quickly separated from the system.
In most companies, something slips quarterly. Product launches will be delayed, sales targets will be missed, and major campaigns will be unperformed. When that happens, most teams readjust and move on.
Finances aren't like that.
Funds are where all these slips are caught, simulated, explained and absorbed. This is the only function designed to see how one team's decisions affect the predictions of another team. This is to tell you where the model was snapped and how it should be rerouted. Without that role, the entire system will begin to drift.
When finances go backwards
Every good finance team builds a kind of muscle memory over time.
I remember which quarter was messy. I know which metrics always work hot. Develop a sense that if 8% growth means actual progression (and not).
If you're too automatic, you lose that intuition. You lose the context behind the numbers and the reflexes asking if something makes sense.
Dashboards continuously update themselves, metrics continue to move, decisions begin to be made, and no one is made without challenges.
It's worth automating
Automation is not an issue. I don't know where to apply it.
AI should speed up the work without replacing it thought Behind that.
- Automate: Vendor matching, workflow routing, forecast generation, and report formats.
- protect: Scenario modeling, capital allocation, trade-off arbitration, judgment, intuition, and narrative.
Our best finance teams use AI every day. They use it:
- It expresses what they may have missed.
- Save time for boring things.
- Ask a better, deeper question.
But they keep the human in a loop:
- All important decisions get a second look.
- If the model cannot explain itself, it will not be shipped.
- It always owns assumptions, not systems.
The goal is to make human judgments and intuition more sharp.
Finance gives you a sense of the outcome of each business decision. That's where your company reasones about the future. It is also a place where cluttered, inconsistent, inconsistent, inconsistent inputs from the entire organization reconcile into one consistent story.
Creating this kind of shared understanding and intuition is far beyond predicting the future.
It cannot be automated that Get away.
If you cut part of your company Please think about it And modeling how everything connects, you become less efficient. You'll get lost.
