Skip to content
Fast Company Executive Board

The Fast Company Executive Board is a private, fee-based network of influential leaders, experts, executives, and entrepreneurs who share their insights with our audience.

Self-driving enterprise still needs a steering wheel

AI can automate the road ahead, but enterprises still need humans to take the wheel when it matters most.

[Photo: Getty Images]

Profile photo of Chris Sluty

Co-Founder and Chief Product Officer · FloQast

As a Tesla owner, I love the promise of self-driving technology. But even as the software becomes more sophisticated, I have no desire to own a car without a steering wheel. Why? Because there are moments—a sudden construction zone, an erratic driver, a patch of black ice—where I need to take control. The same holds true for the self-driving enterprise.

Enterprise technology conversations are dominated by the promise of autonomous agents, leading some to claim that SaaS is dead, and that we are all moving toward a purely AI-driven future.

But as someone who spent years in the trenches of public accounting before moving into product development, I see a different reality. The best of both worlds isn’t found in choosing between software and intelligence; it’s in the confluence of the two. In high-stakes environments, where a single decimal point error can lead to a regulatory nightmare, the goal isn’t total autonomy. The goal is trust. And trust requires a steering wheel.

We are moving toward a world of autonomous agents, but for those agents to be adopted in professions like accounting and finance, they must be built with a manual override.

BEYOND THE “BLACK BOX”

The primary hurdle to AI adoption today isn’t a lack of curiosity; it’s a lack of visibility. Many AI tools currently on the market are built by engineers, for engineers. They often rely on chat-based interfaces that, while impressive for generating a recipe or summarizing a generic email, aren’t the natural place where an accountant goes to work at the end of the day.

Financial professionals don’t just want an answer; they need an audit trail. In our own product development, we shifted away from a purely linear, chat-based model to a visual canvas. This allows users to see the entire flowchart of a process. By visualizing the decision tree, the AI moves from being a “black box” to a transparent partner. It allows the professional to see exactly where the AI is making a turn and, more importantly, where they need to step in.

THE 90/10 RULE: MANAGEMENT BY EXCEPTION

The true value of enterprise-level AI isn’t in replacing humans, but in shifting the human’s role. We should be aiming for AI to handle 80% to 90% of standard, routine transactions. This is the straight road driving. However, the future of accounting isn’t just automation; it’s management by exception.

In any business, the standard operating procedures (SOPs) are easy to document. The challenge lies in the non-standard items, the tribal knowledge that lives only in the heads of senior practitioners. I remember discussing this with my cofounder, Mike Whitmire. During his time as a senior accountant, his team’s biggest customer had a contract filled with non-standard clauses that didn’t fit the typical revenue recognition file. You couldn’t automate that; you had to treat it as a completely different workstream.

AI agents need to be trained to recognize these moments. Instead of guessing when it encounters an anomaly, a well-designed agent should flag the user: “I’ve never seen a contract structure like this before. How should I handle it?” This is the steering wheel moment. By automating the mundane, we give practitioners the space to focus on these high-value exceptions, the office opening in a new country, the complex merger, or the shifting pricing model.

THE PROVE IT ERA OF AI

We are moving out of the era of the flashy AI demo and into the prove it era. Skeptical C-suite leaders no longer want to see what AI can do with a clean, curated dataset in a controlled environment. They want to see what it can do with their messy, real-world data.

To lower the bar for adoption, we have to improve time-to-value. One way we are doing this is through auto-build features that can take a company’s existing inputs and outputs and reverse-engineer the workflow. This changes the sales conversation from, “Trust me, this works” to “Give me your pain points, give me your files, and let me show you the workflow the AI can build for you right now.”

When you can quickly prove value using a customer’s actual information, the skepticism begins to melt away. The customer feels they aren’t just buying a buzzword; they are investing in a tool that understands the unique nuances of their business.

UPSKILL FOR THE HYBRID FUTURE

As leaders, our role in this transition is to foster a culture of curiosity. We shouldn’t mandate AI from the top down; we should highlight the employees who are diving in, finding practical use cases, and getting good outcomes. These individuals are the models for the employee of the future.

In the age of AI, the most valuable skills aren’t just technical; they are cross-functional. We need people who look outside their job descriptions and have sympathy for how their work impacts other departments. The best outcomes happen when sales, marketing, and accounting operate in confluence, and AI can be the bridge that connects those dots, provided we have the documentation and the “single source of truth” to feed it.

The future of software isn’t just AI or just SaaS. It is a hybrid: SaaS providing the guardrails, rules, and structure we’ve relied on for decades, and AI providing the intelligence to navigate through them. We are building the self-driving enterprise. But we’re keeping the steering wheel exactly where it belongs: in the hands of the expert.

Chris Sluty is cofounder and chief product officer at FloQast.