BOSTON — Business leaders are looking beyond a single-minded focus on artificial intelligence and the returns generated by individual agents as they confront the broader challenge of continuously reinventing their organizations around what emerging technologies make possible.
This shifting mindset emerged during CIO 100 Leadership Live Boston, where discussions began with the human dimensions of transformation and quickly moved into the dollars and cents of technology investments. The day ultimately revealed a deeper conversation about how enterprises must continually rethink processes, operating models, technology architectures and investment priorities as capabilities evolve.
Panelists, and an extremely active group of audience participants, presented perspectives increasingly focused on what businesses should become.
It represents a major evolution from the broad calls for “tokenmaxxing” that dominated the AI narrative as we closed out 2025. The idea of setting ambitious transformation goals followed by a disciplined assessment of emerging technological capabilities has fundamentally changed the context of AI agents, automation and emerging autonomous systems. In the process, agentic initiatives, large language models (LLMs), and enterprise and small language models (SLMs) are now seen as the means to a more sustainable end by becoming components of broader enterprise reinvention strategies.
Discussions on return on investment similarly moved beyond measuring the productivity generated by individual tools toward determining whether technology investments systematically and comprehensively improve revenue, customer experience, operating efficiency, resilience or other strategic business outcomes.
Transformation starts with people
The strategic shape of day’s discussions began taking form during the opening session, “Real Leaders, Real Challenges: Championing Tech-Enabled Business Transformation,” featuring Mona Bates, senior vice president and chief information and digital officer at BAE Systems; Sal Companieh, chief information and digital officer at Cushman & Wakefield; and Paul Hlivko, CIO at Optum Financial. Dan Roberts, host of the Tech Whisperers Podcast, moderated the session.
Hlivko described AI as the early stage of a potentially long technology cycle and suggested that leaders should focus on increasing the speed at which their organizations learn rather than simply counting the number of applications moved into production. He compared emerging AI investments with venture capital portfolios, where leaders make multiple bets knowing some will fail while others could generate disproportionate returns.
The discussion nevertheless focused heavily on people. Bates described the CIO as occupying an increasingly important position at the intersection of business and technology. Companieh emphasized frequent communication and creating environments in which employees feel comfortable acknowledging uncertainty. In this context, Hlivko summarized today’s leadership imperative around providing their teams with clarity, culture and conviction in a climate of uncertainty.
Bates added adaptability to the leadership equation, arguing for what she described as AQ alongside traditional measures of IQ and emotional intelligence.
Effective transformation, the panel agreed, will depend first on creating organizations capable of absorbing change.
Complexity becomes a business problem
In a fireside chat titled “The Foundation for AI: Building an Enterprise Ready to Scale,” Jim Chilton, executive vice president of enterprise ventures at Southern New Hampshire University and author of “Enterprise Coherence in the Age of AI,” explored how accumulated technology and legacy process complexity can inhibit the accomplishment of important business transformation objectives.
Chilton argued that complexity rarely results from a single bad decision. It accumulates as organizations add applications, vendors and processes over time to address legitimate requirements as they emerged. AI, he said, requires leaders to expose those layers because deploying agents across fragmented systems requires organizations to understand how information and work actually move through the enterprise.
This, in turn, requires an intimate understanding of all the processes that underpin critical operations. Chilton distinguished between standardized activities that can be automated and differentiating capabilities that may be better augmented by AI. The decision centers on where automation or augmentation contributes to the way an organization creates value.
In “Where AI Delivers: Strategy, Architecture and the Path to the Intelligent Enterprise,” PwC’s Stephen Coyle and Robbie Voigtmann joined Boston Scientific’s Shoubhik Sinha in extending the question of “core value” into enterprise architecture.
Their discussion focused on moving beyond disconnected AI tools and proofs of concept toward reusable capabilities that support production deployments across the enterprise. That requires business, technology and risk leaders to identify use cases according to value, feasibility, risk and their potential to be reused elsewhere.
Companieh illustrated how that reuse plays out in practice when she returned for the afternoon session on “Build, Buy, or Partner: A CIO’s Framework for Scaling AI Across the Enterprise.” She walked through the framework Cushman & Wakefield uses to decide whether to buy a point tool, build in-house or partner for a given AI investment.
She described how a single use case in lease extraction became the seed of an enterprise-wide data foundation. Companieh credited clear roles and shared ownership between Cushman & Wakefield and its technology partner, Unframe, with converting a fast-moving product into measurable business impact.
The push toward reuse, she explained, also exposes how executives can look at increasingly sophisticated AI demonstrations and conclude that their organizations are ready to scale, while technology teams see fragmented data, redundant applications, governance gaps and architectural dependencies that must be addressed.
Transformation confronts the dollars and cents
Those architectural questions led into the session on “Return on Transformation: Time, Talent and Tradeoffs,” where the conversation explicitly addressed the new economics of strategic planning and execution. The discussion featured Lesley Dickson of VantagePoint Strategic Partners, MGX Beverage Group CIO Scott Gardner and Afshean Talasaz, former senior vice president of strategic projects and innovation and chief technology and data officer at Colonial Pipeline.
The panel challenged the idea that organizations should evaluate transformation primarily by measuring individual AI initiatives. The more consequential question, they all agreed, is whether the enterprise itself is becoming more competitive, productive, responsive and/or resilient.
As a result, true transformation begins with defining strategic business outcomes before committing resources to technology, and then assessing how existing tools, available resources and emerging innovations can optimize the chances of success. This last part, they emphasized, is critical, given that the vast majority of large scale transformation initiatives (including mergers and acquisitions) fail to meet their stated objectives.
The panelists warned against automating legacy processes simply because AI makes automation possible. Poor data, unnecessary work and outdated processes do not necessarily improve when machines execute them faster. Organizations, they suggested, should simplify and redesign operations before deciding what to automate.
The same discipline applies to investment. Leaders, they advised, should establish ambitious enterprise goals while making smaller, testable bets underneath them. Successful initiatives can receive additional resources, while those failing to produce results can be terminated.
That approach also requires deciding what not to transform. Some processes warrant fundamental reinvention. Others may require incremental improvement. Many others will likely derive little strategic value from the application of AI in any form.
The result is a broader definition of return. Productivity remains important, but ROI can also include contribution to revenue growth, risk reduction, operating leverage and improved customer experience. In short, CEOs and boards should establish realistic expectations about the economic path from experimentation to enterprise maturity.
Failure becomes something to contain
The afternoon session “Running Enterprise Technology When Failure Isn’t an Option” brought to the stage Paul Beswick, senior vice president and chief information and operations officer at Marsh McLennan, and Sri Sriraman, chief technology officer at Mass General Brigham. The session was moderated by Joan Goodchild, a contributor to CIO and CSO.
The panel kicked off by noting the inherent flaw in the session title. With breaches and incidents occurring every day across a threat landscape that continues to worsen, the leadership team would be well advised to focus on preventing individual failures from becoming catastrophic.
As autonomous systems enter production, it is a task easier said than done. Uncoordinated pilots and vendor agents can create fragmented workflows and dependencies that are difficult to observe or control. Enterprise deployment therefore requires governance, observability and mechanisms that allow organizations to intervene when autonomous systems behave unexpectedly.
Resilience and trusted data become equally important, according to the panelists. High-stakes systems require reliable infrastructure and information foundations calibrated to the risk and value of the functions, assets and technologies they support. The amount of autonomy granted to AI should reflect the consequences should things go wrong at different points on that spectrum.
CIOs begin thinking like investors
The session, “Think Like a VC: Investment Shifts Towards Focused AI Applications,” returned to an idea introduced during the opening session.
Rudina Seseri, founder and managing partner of Glasswing Ventures, and Chris Gardner, general partner at Underscore VC, explored how venture investment practices could apply to enterprise technology decisions.
Seseri described an investment environment focused on identifying sustainable opportunities in AI-native companies serving enterprise, science and cybersecurity markets.
Both advised CIOs to replace static notions of future-proofing by embracing the concept of “continuous portfolio management.” Gardner observed that technology leaders increasingly need to identify market signals amid AI hype, evaluate the adaptability of vendor management teams and distinguish defensible enterprise capabilities from impressive demonstrations. Investments can then be governed through milestones that determine whether organizations scale, maintain or exit them.
In the process, enterprises will need to become more comfortable making calculated technology bets while becoming less tolerant of uncontrolled risk inside production environments.
Reinvention becomes continuous
The closing session, “What’s Next for the CIO: Preparing for the Next 12-24 Months,” extended that argument into the future of work.
Boston Dynamics CIO Chad Wright said organizations should begin preparing for three categories of coworkers: humans, virtual agents and physical robots. Boston Dynamics already has its Spot (quadriped) robots working across hundreds of organizations, while humanoid (bi-ped) robots introduce additional questions involving infrastructure, physical safety, accountability and workforce acceptance.
Terry Carpenter, CIO of MIT Lincoln Laboratory, stated that, as agents assume tasks traditionally performed by entry-level employees, organizations will need to seriously consider how younger professionals are trained to develop the judgment, delegation skills and institutional knowledge to move into middle management roles, and ultimately, senior leadership positions.
Jacqui Nevils, Global CIO, Fresenius Medical Care, and a veteran Air Force officer, emphasized that there is a growing case for new and significant investments to be made in onboarding talent early and training them right away for major responsibilities. If the next generation of employees are to supervise multiple agents (along with human subordinates) while remaining accountable for their output, organizations will need to teach skills that were previously acquired through years of progressively more responsible work.
The discussions ultimately brought the program back to where it started.
While the agentic economy is introducing another generation of enterprise technology, it is also accelerating the rate at which organizations must reconsider how work gets done, where capital gets allocated, who does what, and which risks are acceptable as new capabilities are deployed to provide competitive differentiation.