As technologies like AI transform industries overnight and budgets come under scrutiny, leaders face more pressure than ever to separate the truth from the hype and bring the most effective solutions into their organizations. But success is always about more than the specifics of the innovation itself.
The only way to discover how new technology will behave in your environment, and how much your organization will benefit, is to roll up your sleeves and start building your own prototypes.
Fund small experiments. Evaluate swiftly for proof of capabilities so you can move on to the important work of listening, learning and adapting. Most importantly, don’t fear failure. It’s important to be technically fearless and know that you learn as much from failed prototypes as successful ones.
Those prototypes are roadmaps to the future. Our team at Booz Allen saw this firsthand when we built a prototype of an early agentic AI system 18 months ago so we could understand:
-
How agents work together to solve problems; and
-
Loose coupling in distributed systems.
As we dug in, however, what we learned led to something even more valuable: the agentic mesh and AI foundry that now fuel innovation across our business.
Hype is cyclical
As you experiment and procure tech with an eye on the future, don’t overlook lessons from the past. Technology follows patterns. I’ve observed a 20-year oscillation over the course of my career between edge computing and centralization. To “skate where the puck is going,” it’s important to study these cycles.
As an example, latency and bandwidth used to drive an enterprise’s tech decision-making. Leaders worried whether their network could keep up with the pace of innovation at scale within the cloud or a data center. Now, though, an artificial intelligence-radio access network (AI-RAN) has the potential to reshape how organizations experience connectivity by making networks faster, more adaptive and more energy-efficient for analytics, security and other AI-heavy workloads. This is pushing enterprises back to edge computing, as network predictability increases and enterprises realize AI-RAN’s potential as a performance multiplier.
Experiment and procure with resilience built in
All this experimentation and evaluation is for naught, however, if your production systems get knocked down in a cyberattack or lost in a maze of competing systems. If your production system is not secure, reliable, scalable and resilient, none of the features matter.
One way to ensure resilience as you procure tech at the cutting edge is by investing in more than one solution for any critical process or technology. In other words, don’t put all your eggs in one basket. I let the “rule of three” help govern my decisions; investing in no less than three heterogeneous technology solutions to explore, imploring my teams to prepare for at least three scenarios, etc.
As the leading provider of AI to the federal government, we constantly evaluate and re-evaluate our technology. At this time in AI-driven software development, I initially made the choice to invest in seven different tools. However, as developer feedback and usage data determine what is working, within a finite time window — six months, max — we will streamline our investments to keep pace with the AI advancements that will have the largest impact on our business and customers.
Tech isn’t the bottleneck, people are
While these principles of building and buying new technology are crucial, adoption bottlenecks often stem from factors such as employees’ perceptions of control and their ability to do their jobs.
I saw these dynamics in action at the Pentagon during the adoption of Wi-Fi. The mission needed it for the rise of the laptop and mobile era, yet many wanted to ban it outright, considering it inherently unsafe and operationally risky. We changed their mindset by addressing these security concerns head-on and demonstrating that Wi-Fi could be hardened, monitored and governed.
Reframing the conversation is just one of many tech-focused change management tactics:
-
Build and prepare your team. Don’t focus on only technical brilliance. Find individuals who have operational and process expertise. When onboarding, set cultural norms and expectations early so everyone is working out of the same playbook.
-
Present a consistent leadership style. My own “10 Leadership Rules” — the principles I lead by no matter the job and developed over the course of my decades-long career in tech — have proven invaluable for establishing cultural norms and teaching teams how to think, not just what to do.
-
Evaluate your approach to collaboration. Regular feedback loops and whiteboarding sessions can be valuable for drawing conversations out of disparate silos and into the light, where they can guide and accelerate progress. Build these into your processes and test from the get-go.
-
Rethink traditional roles. We’re seeing this in action right now through our implementation of spec-driven development. The efficiencies gained and time saved free capacity for experimentation. But there’s a learning curve involved in developing these specifications, and a fundamental shift in roles as developers become AI managers. As a leader, be willing to take the time required to guide your team through such evolutions.
The leaders who best meet this current moment are those who anticipate the cyclical nature of evolution; are willing to experiment, build optionality, feedback, and resilience into their systems; and bring their people along with them.
