The automotive industry is entering one of the most disruptive decades in its history—not just from electrification and shifting consumer expectations, but now from the rise of agentic AI.

A new McKinsey report, analysing over 50 implementations, shows why most companies struggle to capture real value from this technology—and how leaders are pulling ahead.

For automotive founders, OEM executives, and investors, the message is clear: the winners won’t be those who chase flashy AI demos. They’ll be the ones who reimagine workflows, build trust into every deployment, and design for scale across complex global operations. These six lessons show exactly how.

1. Fix Workflows, Not Just Agents

The most significant failure point? Companies often obsess over the AI agent itself, while overlooking the broader operational context. Successful deployments reimagine entire workflows—how people, processes, and technology connect.

In the automotive industry, this matters more than ever. Think about warranty claims, parts logistics, or customer service—dropping an AI agent into a broken process won’t create value. But redesigning the workflow around seamless human-AI collaboration can.

One legal services provider in McKinsey’s research created continuous learning loops where every user's edit improved the system. Imagine the same principle applied to connected vehicle diagnostics—each human correction feeding back into smarter, faster AI-driven service.

2. Agents Aren’t Always the Answer

Not every problem needs an agent. McKinsey found that highly standardised, low-variance processes—like invoicing or demand forecasting—often benefit more from traditional automation or predictive analytics.

Automotive leaders should reserve agentic AI for areas where variability and judgment truly matter, such as customer interactions, complex problem-solving, and creative engineering scenarios. That’s where agents can unlock true competitive advantage.

3. Quality and Trust Are Non-Negotiable

The report warns against “AI slop”—impressive demos that collapse in real-world use. In a safety-critical industry like the automotive sector, this risk is amplified. A system that works 80% of the time isn’t just frustrating—it’s unacceptable.

The best companies treat AI like a new hire: give it a job description, onboard it properly, and provide continuous feedback. A global bank identified every mismatch between AI recommendations and human judgment, then refined the system until trust was built.

Automotive firms must apply the same discipline, with rigorous evaluation frameworks and feedback loops. Trust is the foundation for adoption—and adoption is where the value comes from.

4. Build Once, Scale Everywhere

One of the most practical insights: avoid creating one-off agents for every task. Leaders build reusable components and centralised platforms that can be deployed across multiple workflows.

This cuts 30–50% of redundant development work while ensuring consistency. For global automotive companies spanning markets, product lines, and service networks, this is the difference between scale and expensive chaos.

And don’t forget monitoring—when something goes wrong, you need granular visibility into which workflow step is causing the issue.

5. Design for Human-AI Collaboration

The goal isn’t to replace humans—it’s to redesign work so people and AI collaborate effectively.

Agents are excellent at handling routine tasks, but humans remain essential for making judgment calls, addressing edge cases, and ensuring accountability. In one insurance example, 95% of users accepted AI recommendations because they were transparent and easy to act on.

Automotive leaders should take note. Whether it’s a dealer explaining a repair, or an engineer resolving a design anomaly, collaborative AI tools can enhance—not diminish—human performance.

6. Focus on Reusable Value

The final trap? Companies rush to build agents for every identified task, creating waste. Winners focus on recurring patterns and build reusable capabilities that can be applied across multiple contexts.

For the automotive industry, this could mean building a claims-processing engine that works across geographies, or a customer-interaction layer that scales from sales to after-sales care. The point: don’t create scattershot solutions. Build platforms that deliver compounding returns.

The Strategic Opportunity

These six lessons explain why some companies capture disproportionate value from agentic AI while others end up with expensive experiments that never scale.

For the automotive industry, the opportunities are immense—streamlining warranty claims, optimising global supply chains, building safer cars, and transforming customer experiences. But success requires discipline: workflow-first design, rigorous quality control, scalable architecture, and human-AI collaboration.

The race isn’t about who has the most AI—it’s about who has AI that works reliably, scales efficiently, and delivers measurable customer value.

Now is the time for every automotive leader to audit workflows, identify where human-AI collaboration creates immediate impact, and invest in platforms that scale. Those who act decisively today won’t just capture value from agentic AI—they’ll shape the future of the industry.

Have a great week!