Earlier this week, I joined three fellow industry CEOs at the EFCG CEO Conference for a discussion on “The CEO’s AI Moment: Lead It or Lose Ground.” Together, we examined how our firms are creating opportunity through high-impact technology and AI adoption — from the organizational changes and investments required to the resistance encountered and lessons learned along the way. At the heart of the conversation was a question every industry leader must consider: What will separate the firms that capture the potential of AI and emerging technologies from those that fall behind?
Photo Credit: EFCG
At Atlas, those priorities are shaping how we adopt AI across the enterprise. Our objective is to improve how work is performed while keeping human expertise and judgment at the center. That begins with executive ownership. My view is straightforward: the CEO must own the work personally rather than delegate it.
That ownership matters at Atlas. We are an infrastructure and environmental firm with approximately 3,100 employees across 80 U.S. locations, built initially through 15 acquisitions. Most of our work is field-services oriented, spanning construction engineering and management; testing, inspection and certification; remediation management; permitting and compliance; and healthy buildings analysis.
We have built an enterprise-wide approach to AI adoption designed for that operating reality. I set the initial direction and worked with our executive leadership team to establish the policies and expectations that would guide adoption across the organization. Five elements have defined our progress to date.
1. Executive Ownership Sets the Pace
Over the past 18 months, I have deliberately increased my proficiency in using AI. That direct experience has been critical to understanding both its benefits and its risks — and to shaping an AI strategy, governance structure, business case and usage policies suited to Atlas.
I established the same expectation for our executive leadership team: become proficient users, with every leader actively using AI within three months. Training, coaching and small-group experimentation supported that commitment.
The principle is simple: leadership leads and the organization follows. In my career, few examples have demonstrated more clearly how quickly leadership behavior can create an outsized impact across a company.
2. Governance Turns Ambition into Action
Enterprise adoption requires more than access. We established a companywide AI policy with clear guardrails and expectations, centered on making governed AI tools available to every Atlas employee. Our bespoke systems are built in-house with no external data sharing, and every tool or agent must demonstrate a form of return on investment before deployment.
We also invested in the leadership and organizational structure required to sustain the effort. A Senior Director of Enterprise AI oversees the work, supported by AI Champions across the company and within each business unit and function. Deliberately, that champion network is not a top-down mandate — an important distinction in a workforce with this many hourly and field employees.
Our AI Pods bring that structure directly into the business. These cross-functional teams of three to five AI specialists work alongside technical and field staff on tool development and enablement. Through ride-alongs with field and office employees, they gain a firsthand understanding of real problems so that the solutions we build remain close to the work and are actually used.
3. Adoption Must Reflect How the Workforce Operates
Approximately 65% of our employees are hourly. Because Atlas is primarily field-services oriented, most are not desk-based or using prompts in their daily work. That reality changes the model for implementation and becomes an advantage once the organization solves for it.
Our implementation therefore extends across both internal operations and client-facing work: finance and accounting, marketing and proposals, HR service enablement, contract review, project planning, workflow automation, field tool development, reporting, data mining and evaluation, project management, scheduling and safety.
4. Redesign the Work Before Deploying the Technology
A significant share of this initiative has focused on the foundation beneath the technology: re-evaluating and revising business processes while consolidating data from Atlas’ acquired companies into a common format. After 15 acquisitions, our data was, as I describe it, “everywhere and nowhere.”
Meaningful AI deployment first required us to build a unified data lake and reengineer processes, with AI positioned on top of both. Given our acquisition history, this has been a substantial undertaking. The objective is to make 20 years of accumulated data usable by AI systems going forward.
I have seen organizations deploy AI agents without first redesigning the workflows in which those agents operate. The result is often faster execution of an existing process — not a meaningfully better process. Our ongoing redesign work has surfaced and corrected inefficiencies that existed independently of AI. By changing how the work itself gets done, we can pair AI adoption with improved execution rather than simply accelerate yesterday’s work or bolt new technology onto broken or inefficient processes.
5. Human Judgment Is Nonnegotiable
Human review of every AI input and output is a firm requirement at Atlas. Two concerns underpin that standard: the erosion of independent reasoning when employees accept AI-generated output without applying their own expertise and judgment — what I call cognitive atrophy — and the proliferation of low-value output, or “slop grenades,” that leaves others to spend additional time filtering for what is relevant.
For that reason, Atlas has no fully autonomous AI applications. People make the final call on inputs and outputs. Our governance is rigid, and everything is hosted within the company. Many employees already use AI to execute tasks and gain efficiency, but that is not where the opportunity ends.
That standard of human oversight also shapes where we go next. As AI advances across Atlas, the opportunity is to move beyond task execution and efficiency and develop it as a thought partner and upskiller — one that builds and challenges thinking rather than outsourcing it.
The firms that realize the potential of AI will be those that treat adoption as a leadership commitment, not a technology rollout. At Atlas, the five pillars give us a disciplined foundation for that work and a clear path forward. The technology will continue to evolve. Our responsibility as leaders is to evolve with it — thoughtfully, deliberately and with the judgment to turn possibility into lasting value.