There Are No Rollbacks In The Arctic: Leadership Lessons For Building Under Pressure In The AI Era
Shubha Govil is Chief Product Officer at Sauce Labs, where she leads product strategy and innovation at the intersection of AI and SDLC.
gettyWhile walking through the uneven terrain of Auyuittuq National Park in the Arctic for 11 days, I could only focus on one step at a time. The Arctic can be an unforgiving environment, and any failure could be the difference between life and death. Whatever you carry is what you have to work with. If you forget fuel, which we almost did, you might have to survive on snack bars and cold food. This is an environment that offered no rollback, no system retry and no second replica standing by in another region.
Being in the software development industry for over 20 years, I’ve gotten so used to the systems that are designed to be redundant to address failures seamlessly and reliably. Now, the gold rush of AI is making it harder to maintain the same level of rigor to enable fault tolerance and keep the system blast radius bounded. Technology leaders are also under unprecedented pressure to deliver with speed and with little margin for error.
Here are my learnings from an environment where having no redundancy turned out to be the most useful leadership lesson I’ve learned.
Precision is a word often synonymous with slow. On an expedition, however, the fastest teams are the ones that don’t have to stop continuously to fix something. Precision is how they get their speed. It’s about consistency in your craft and reducing the amount of random errors, be it hiking or how we deliver products. Every consistent step moves you forward and builds the pace.
It’s not much different in software development. You have to execute consistently again and again. It requires excellence not in just one area or by one team but across a full life cycle of delivery. As AI-driven productivity is a default expectation, delivering value requires human precision to close the gap between improvements brought by AI and non-AI systems.
Our trek didn’t require ropes to line up for a climb, but we had to move forward as a team with the pace that allowed enough room for emergency delays. If one person was slow, the guides made sure they set a pace that not only kept us moving but worked for all.
A common failure mode I see in many organizations as teams adopt AI-native SDLC is that not all functions are accelerated at the same pace. AI-generated code accounts for up to 75% of all new code, but quality, security, support and GTM have not yet caught up. Every function is getting faster individually, but as a whole, the system is slowing down and carrying more risk.
Leaders expecting velocity need to walk each step together with their teams and ensure end-to-end productivity gains rather than settling for heroics in functional silos.
Everyone who walks long distances knows that the most painful reason for slowing down comes from unattended blisters, which become increasingly painful with each step if not addressed early. However, most people still don’t stop. The reason is mostly social, as nobody wants to be the person to slow down the group.
That’s precisely the dynamic in many teams where people continue to let each point of friction grow, as it’s uncomfortable to stop to fix a blocker others don’t see. Over time, multiple points of friction cause gradual degradation of excellence. This leads to shortfalls that show up at the worst possible moment when delivery dates are committed to customers. Leaders need to empower teams to quickly name these friction points and normalize addressing rising hotspots.
I didn’t sign up for this trip initially because I believed it was beyond my limit. As it turns out, I did it, and it proved that my belief was an assumption rather than a fact. It had been set by an earlier version of my perception of myself and what I’m capable of achieving. It was only when I actually pushed myself to make it happen that I realized how wrong my beliefs were.
The limit is always further out than we imagine. If you set your mind to a goal and bring passion, will power and adaptability, they’ll carry you past the point where you expected to stop.
Teams also carry many assumptions and beliefs about their limits. One such assumption that teams must reevaluate now is whether they can drive productivity gains along with system reliability. The limit is further out when the first three factors are in place.
In the software development industry, the potential that AI offers is huge—but only if we believe that humans are capable of pushing past our current limits to solve the right problems with the right guardrails and governance.
Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?
