Four minutes.That is how long the pilots actually fly the airplane by hand on a typical commercial flight. Cranfield University analyzed 14,000 flights on Airbus A319 aircraft and found that nearly 80% of those flights had just four minutes of manual flying time. Autopilot handled roughly 90% of the journey.*
Most people hear that and think the pilots are not needed.
The opposite is true. Those four minutes of hands-on flying are the highest-stakes moments of the entire flight, and the other 96% of the time the pilots are not sitting idle. They are doing the work that actually keeps 200 people alive at 500 miles per hour. And that distinction is the single most important thing to understand about AI in your business right now.
Let me explain.Last month, I was in seat 18C on American Airlines flight 1534 from Philadelphia to Denver, watching the Rockies come into view through a wall of wind (my contact on the ground in Denver later told me he was not surprised we were delayed, given how hard it was blowing that day), and I started mapping the parallels between what those pilots do and how we use AI in our engineering practice.
The pilots had spent 45 minutes before we even pushed back from the gate doing work that most passengers never see. Filing the flight plan, reviewing weather data and NOTAMs, calculating fuel loads and weight distribution, running preflight checklists system by system to confirm every instrument and sensor was ready. They coordinated with ground control, briefed each other on the departure procedure, and confirmed their contingency plans for anything that might go sideways.
Then came taxi and takeoff: 100% manual, 100% human judgment. Every single commercial takeoff is flown by hand. There is no such thing as an automatic takeoff anywhere in commercial aviation. That requires split-second decision-making, situational awareness, and physical skill that no automation can replicate today.Somewhere around 1,000 feet above the ground, the autopilot engaged.
For the next three and a half hours across Pennsylvania, Ohio, Indiana, Nebraska, and Colorado, the airplane largely flew itself. But the pilots were monitoring systems, communicating with air traffic control, adjusting the flight management computer as conditions changed, watching weather radar, coordinating with dispatch, and maintaining the situational awareness needed to intervene the instant something required human judgment. The autopilot handled the repetitive, steady-state task of maintaining altitude, heading, and speed. The pilots focused on the high-value, high-judgment work that only trained humans can do.
Then, as we descended into Denver through that wind, they took back full manual control for the approach, landing, taxi to the gate, and postflight procedures. Less than 1% of commercial landings are fully automatic.** The pilots finish every flight the same way they started it: hands on, fully engaged, applying years of training and experience to the moments that matter most.
This Is Exactly How AI Works Best in Your BusinessWhen I talk with engineering firm owners and technical leaders about AI, I use this analogy constantly because it maps so cleanly to reality.Think about how most engineering projects actually work. There is a front-end phase where humans need to be deeply involved: scoping the work, understanding the client’s needs, assembling the right team, developing the technical approach, and setting up the “flight plan” for the project. This requires engineering judgment, client relationship skills, and strategic thinking that no AI can replace.
Then there is the long middle stretch. The documentation, the data entry, the formatting, the repetitive calculations, the status reports, the email drafts, the meeting summaries, the progress tracking. This is the cruise altitude of your project, and it is where most engineering teams spend 60% or more of their time on work that does not require their full expertise.
And then there is the back end: final review, quality assurance, client delivery, lessons learned, and closing out the project. Again, this is hands-on, high-judgment work where human expertise and client relationships are essential.
The pattern is the same. Humans lead the setup and the finish. Automation handles the steady-state middle. And humans stay in the loop throughout, monitoring and intervening when judgment is needed.
What Changes When You Start Flying with AIIn my own engineering practice, we have applied this exact framework across dozens of workflows. The results are consistent and measurable.Proposal development that used to consume 15 to 20 hours of senior engineer time now takes a fraction of that, because AI handles the initial drafting, formatting, and data assembly while our engineers focus on the technical strategy and client-specific customization. Technical documentation that used to be a manual grind now flows through AI-assisted templates that maintain our quality standards while freeing our team to focus on the engineering analysis itself.
One of our accelerator participants built an AI agent to review legal terms and conditions for sales contracts. What used to take 2 to 4 hours of careful line-by-line reading now takes about an hour with a redlined copy ready for human review. The human still makes every final decision. The AI handled the steady-state scanning and flagging.
Another participant automated their sales team’s customer check-in process. The scheduling, drafting, and sending of routine outreach emails, the kind of work that used to pile up and fall behind, now runs smoothly with AI handling the repetitive execution while the team focuses on the actual relationships.In every case, the human is the pilot. AI is the autopilot. And the human never leaves the cockpit.
The Three Phases of Every AI-Enhanced WorkflowWhen I help engineering firms map their workflows for AI integration, we use a simple framework that mirrors the flight analogy:Phase 1: Preflight (Human-Led Setup) Define the objective, gather inputs, set quality standards, and configure the AI tools for the specific task. This is where engineering judgment, client context, and strategic thinking drive every decision. You are filing the flight plan and running your checklists.Phase 2: Cruise (AI-Augmented Execution) Let AI handle the repetitive, time-intensive middle work: drafting, formatting, data processing, initial analysis, scheduling, and routine communications. You monitor the outputs, make course corrections, and intervene when something needs human judgment. The autopilot is engaged, and you are watching the instruments.Phase 3: Approach and Landing (Human-Led Completion) Take back full control for final review, quality assurance, client-facing deliverables, and strategic decisions. Apply your expertise to the moments that matter most. Then complete your postflight: document what worked, refine your processes, and prepare for the next project.
Why This Framing Matters for Your TeamOne of the biggest barriers to AI adoption in engineering firms is the fear that AI replaces human expertise. That fear is understandable, but it is based on a misunderstanding of how AI actually works in practice.
Pilots are not less important because autopilot exists. They are more effective. Autopilot frees them from the cognitive load of maintaining a constant heading and altitude so they can focus their attention on navigation decisions, weather avoidance, communication, and emergency preparedness. The result is safer flights, more consistent operations, and better outcomes for everyone on board.
The same is true for your engineering team. When AI handles the repetitive documentation, data processing, and routine communications, your engineers get to spend more time on the high-value analysis, creative problem-solving, and client relationships that actually drive your business forward. They do not become less important. They become more focused on the work that only they can do.
After implementing AI across 40+ engineering projects, I have seen this pattern consistently: teams that embrace the pilot-and-autopilot model do not just save time (though they do, typically 3 to 8 hours per person per week). They also report higher job satisfaction because they are spending more of their day on meaningful engineering work instead of administrative tasks.
The Bottom LineYou would never board a plane without pilots in the cockpit. And no responsible airline would send a crew up without autopilot in an aircraft designed to use it.The combination of human expertise and intelligent automation is what makes modern aviation the safest form of transportation in human history. The same principle applies to your engineering practice.
You are the pilot. Your AI tools are the autopilot. Together, you cover more ground, deliver more consistent results, and free up capacity for the work that truly requires your expertise.
The firms that figure this out first will have a measurable advantage in capacity, quality, and team retention. The ones that keep hand-flying every minute of every project will burn out their best people and fall behind.
After implementing AI across 40+ engineering projects, I can tell you: the framework works. We test everything in our own practice first, and the pilot-and-autopilot model is how we approach every workflow.
Start building your flight plan today. Message me “PILOT” and I will send you our Engineering Acceleration AI Roadmap, a free tool that helps you identify exactly where AI can save your team hours every week.
*Source: Cranfield University Safety & Accident Investigation Centre
**Source: FlightDeckFriend.com
