Self-driving cars and automated decision-making in business have more in common than it seems. Both evolve step by step—through alerts, assisted decisions, and eventually, full autonomy. Understanding this journey is key to scaling smarter, faster, and more impactful business decisions.

When I was a child, I thought cars would drive themselves. And I loved the idea. Sci-fi books from the ’60s and ’70s reinforced this idea, and I was obsessed with the possibility of being seated, relaxed, in the co-pilot seat. I was sure that, from one day to the next, someone would make them a reality.    

Well, here we are in 2025, and cars are still not fully autonomous. Even though there are around 1.5 billion cars in the world, the number of 100% autonomous commercial cars remains roughly zero today.    

Nevertheless, cars have improved to automate certain actions over the last 30 years or more. While cars are not yet fully autonomous, most cars we drive today brake automatically or warn us with lights and alarms when they detect an impending crash. Automation is not a breakthrough; it is a process that takes years or even decades. However, Elon Musk estimates that in five years, 50% of the vehicle fleet will be autonomous, and in 10 years, no one will have to lift a finger to drive.   

What Self-Driving Cars Can Teach Us About Automating Decisions 

Time went by, and through my life, choices, and work, I ironically ended up focusing on something as complex as autonomous cars: autonomous business decisions. 

Just as self-driving vehicles did not emerge overnight, neither did decision automation in companies. Over the last years, I have worked with organizations to help them identify where automation creates value, align it with business goals, and scale the impact of smarter decisions.    

Automated decision-making in business is happening gradually—and that is a good thing. Rushing a decision can be costly. In this process, I have learned—through my work with some of the world’s largest companies—that automation does not happen all at once. It starts with alerts that blink when something is off. Then systems begin to act—unless we tell them not to. Soon, they start recommending what to do. Then, they act on their own and inform us subsequently. And finally, they stop asking altogether—because they have learned how to decide.    

Whether we are talking about cars or business decisions, these are the five stages: 

1) Only Automated Alerts 

Modern dashboards look like Christmas trees: low tire pressure, fuel level, object detected… These alerts help drivers focus on what matters. This is the lowest automation level, leaving all decision power to humans.    

In business, this level of automation is called “Decision Support Systems”. They reduce the burden of constant monitoring and provide timely, relevant signals. But the key is not to collect all the data—it is to focus on the important decisions that matter most, those that can put our business at risk or make a relevant difference.    

That is the mindset shift: from data-driven to decision-driven. The right alert is not the one with the most data, but the one that supports a meaningful choice. 

Think of a sales team that receives alerts when inventory of a high-demand SKU falls below a critical threshold. No more spreadsheets, no more fire drills—just a timely signal, so action can happen before it is too late. 

2) Semi-Automated with Human Override 

Ever had a car ask if you want to parallel park—and then start doing it unless you intervene? That is how many workflows function: the system is ready to act unless someone steps in.    

That was the approach in a tool we helped design for Key Account Management. RPA consolidated data from multiple sources and validated the status of each account. Based on this, the system automatically recommended renewal and investment levels—moving forward unless someone chose to adjust. The system made the first move; the team could step in, review, and redirect as needed.    

These “cancelable automations” save time while still giving people control when it matters. Leaders become conductors of complex systems—choosing when to let automation run and when to intervene. 

3) Augmented Decisions

If we add one more level of automation, we reach “Augmented Decisions”, where humans and analytics decide together.    

Picture this: You are driving on a gravel road. It has been raining for 20 minutes, and you are moving over 60 km/h. As the mud thickens, the road twists and steepens. The car does not just react—it analyzes the full context and suggests a plan: reduce speed, adjust traction, lower tire pressure. You are still driving, but you are no longer deciding alone.    

In business, this shift is a game-changer. Machine learning enables systems to move beyond pre-set rules. They learn from patterns, recognize evolving situations, and actively collaborate with humans—recommending, modeling, and prioritizing actions.    

During a project with a leading FMCG company, we trained an algorithm to forecast volume by consumption occasion and align investment accordingly. Machine learning was used to recommend investment levels based on expected revenue per occasion—adapting to HUB strategies while staying aligned with OU priorities. The team remained in control of the final decision but now with sharper insight and greater precision. 

4) Automated Action with Post-Review 

My car once braked suddenly to avoid a collision. I did not ask for it—and honestly, it was a bit frightening. But it probably saved me from a serious collision.    

In organizational terms, this is when the agent executes and informs us afterward. Inventory is redistributed when it hits a critical threshold. A machine shuts down to prevent damage. The system acts in the moment, while review and refinement come later.    

We applied this concept in a project where we automated decisions around cold drink equipment. When usage dropped below a predefined threshold, the system automatically flagged the unit for removal and triggered a replacement order. Teams could later adjust parameters—but the action was already in motion.   

5) Fully Autonomous Decision-Making Systems (ADMS)

We are not quite at Isaac Asimov levels yet, but we are getting closer.    

When talking about cars, even though there are still legal and user habit concerns that need to be solved, Full Self-Driving systems sound technically quite close to complete autonomy.    

In business, we are already seeing fully automated decision-making systems in action—systems that operate with absolutely no human intervention:    

1. Approving loans based on real-time scoring.    

2. Adjusting prices dynamically every few minutes.    

3. Triggering predictive maintenance without alerts.    

4. Reorganizing logistics networks on the fly.    

In pricing, for instance, we developed a system for a major consumer goods company that adjusts promotional discounts in real time based on demand patterns, margins, inventory, and competitor behavior. Once configured, the system runs on its own—freeing commercial teams to focus on higher-level decisions.

These five stages are more than a roadmap—they are a Decision Intelligence journey. One that helps organizations scale maturity over time, connect decisions to outcomes, and turn automated decision-making into a driver of real strategic value. 

How Smarter Decision-Making—Not Just AI—Drives Real Business Transformation 

AI and machine learning are evolving rapidly. Just like driving today feels nothing like it did when I first got behind the wheel, the landscape of decision-making is radically different from when we launched Tandem 18 years ago. 

Back then, we believed better decisions could drive better performance. We are still certain about that; what we did not know was how quickly the world would change—and how essential it would become to keep evolving with it. 

An integrated Decision Intelligence Solution should help organizations identify the decisions that matter most—and then build the systems, data, and logic to automate them where it makes sense. Do not start with the tech; start by understanding the business: what decisions drive results, what data really matters, and how processes can be improved. The goal is to figure out where alerts are useful, when augmented decisions make sense, and when it is time for fully autonomous systems. The aim is not just digital transformation—it is smarter decisions that save time, cut costs, and drive real impact. Because the future of performance is not about having more data. It is about making better decisions. 

Companies today must make more decisions, faster and under greater complexity. We believe that optimizing how those decisions are made is key to improving performance. This is not a shift that happens overnight—nor is it about replacing people. It takes time, analysis, and a willingness to ask the right questions. Among them: Are we headed in the right direction? And which decisions are we ready to entrust to machines? 

That is the road we are on. That is the road many companies are on. Are you on this path? 

Asimov once imagined a future where machines could make decisions for us—guided by clear rules and logic. But even in “I, Robot”, things went wrong when systems acted without context, empathy, or alignment with human goals. 

That message still holds. Technology alone will not improve decisions. What makes the difference is how we design, supervise, and integrate it—so that automation becomes an extension of our judgment, not a replacement for it. 

The future of automated decision-making does not belong to machines. It belongs to the people who know how to use them wisely. 

Gastón Francese

gf@tandemsd.com