Artificial intelligence is forcing CIOs to reconsider the economics of enterprise technology. While application development costs are falling, autonomous systems are complicating governance as traditional productivity measures struggle to capture returns on transformation investments for C-suite executives and boards of directors who are interested in documenting tangible business gains from agentic deployments.
These were among the issues that dominated CIO 100 Leadership Live Chicago on Oct. 13, where enterprise technology executives, industry specialists and investors examined the practical implications of moving AI from experimentation into everyday business operations.
Speakers repeatedly observed that technological innovation is outpacing the ability of traditional enterprise processes to absorb it. The theme surfaced first in the opening panel, “Championing Tech-Enabled Business Transformation,” moderated by Dan Roberts, host of the Tech Whisperers podcast.
Laura Walsh, CIO of Culligan, described the conflicting expectations technology leaders face. They are asked to accelerate AI adoption, demonstrate immediate financial returns, maintain security and quality, and fund new initiatives within existing budgets.
Walsh stressed the importance of established management disciplines, including data governance, project execution and technology business management. She outlined a hub-and-spoke approach that establishes common AI controls while allowing individual business units to experiment.
Jamie Head, founder and CEO of Stack360, contended that CIOs must work more closely with CEOs and CFOs to identify a limited number of strategic priorities and decide which initiatives to defer. Business leaders, he added, increasingly look to automation to improve margins and productivity, creating an imperative for CIOs to play a key role in separating worthwhile investments from technological distractions.
AI economics move beyond productivity
The financial implications of AI adoption surfaced throughout the day, particularly the gap between gains in employee productivity and measurable financial returns.
In a discussion of the intelligent enterprise moderated by Jon Glick, principal and advisory CMO at PwC, PwC principal Scott Behrens said AI transformation requires redesigning business processes from the ground up. Organizations that layer disconnected AI applications onto existing workflows risk increasing technology consumption without corresponding business benefits, he warned.
Meghna Shah, principal for engineering and AI at PwC, pointed to the widening gap between the speed of AI-assisted software development and the engineering discipline required to deploy reliable enterprise applications. AI can dramatically accelerate prototyping, she noted, but organizations must still address architecture, integration, security and quality assurance before those applications can support critical business operations.
Andy Xydakis, global CIO of Aon, said the company has expanded its cloud footprint from about 35% to 85% of its application portfolio while introducing AI capabilities to roughly 60,000 employees. AI-assisted proposal development has cut work that once required 40 to 50 hours to about four or five hours, according to Xydakis. Aon has also strengthened its financial operations capabilities to monitor technology consumption and manage costs as adoption expands.
A subsequent session examined how to secure returns on transformation investments in emerging technologies such as AI. Chris Campbell, CIO of DeVry University, said that his team measures investments against broad institutional objectives, including student enrollment, persistence, retention and graduation rates, in addition to savings from individual technology projects.
Campbell also maintained that CIOs must understand the business capabilities their organizations need before choosing applications or technologies. AI is making software development more accessible, he noted, so technology leaders must work with business stakeholders to set common priorities and guardrails while managing expectations about what emerging technologies can realistically accomplish.
Prashant Shanti Kumar, global head of data domains at Marsh, recommended setting clear business objectives and breaking complex transformation initiatives into manageable phases. Drawing on his experience integrating organizations and business processes, including at a major national insurance carrier, Kumar advocated 90-day implementation plans with measurable outcomes that let teams adjust priorities as conditions change.
Gone, both intimated, are the days of multi-year strategic technology and deployment plans.
Distributed development raises governance challenges
As the economics of enterprise technology grow more complicated as the result of increasing numbers of autonomous agents and applications, Michael Blank, chief operating officer of Retool, said AI is dramatically reducing the cost of generating software while leaving enterprises responsible for the infrastructure, maintenance, security and governance required to run it.
Blank warned that business employees can now build applications without IT’s knowledge, creating operational dependencies and exposing information to poorly governed systems. He recommended allowing development throughout the organization under consistent enterprise controls over identities, permissions, data access and application deployment.
The growing autonomy of AI systems is also forcing cybersecurity leaders to rethink traditional protection strategies, the focus of a security and risk panel moderated by CIO contributor Joan Goodchild.
Heather Ceylan, chief information security officer at Box, said organizations must monitor and restrict what an AI agent actually does, in addition to controlling which resources it can access. Agents may operate with broad permissions, process large volumes of sensitive information and execute actions far faster than human employees, she explained. Ceylan said security controls therefore need to operate during execution, when potentially harmful behavior can be identified and interrupted.
Ceylan described using security agents to review code, perform threat modeling and help enforce engineering requirements as software development accelerates.
Greg Sylvester, vice president of enterprise AI and infrastructure at Cisco, discussed the importance of monitoring agent activity, detecting malicious instructions and building responsible AI requirements directly into enterprise platforms.
Michael Chahino, CIO of Elgin Community College, outlined additional challenges facing higher education, where decentralized decision-making and widespread use of personal AI subscriptions complicate governance. His institution is emphasizing education and cross-functional oversight while developing policies for responsible AI use.
The discussion pointed to a broader need for greater visibility into the behavior of autonomous systems and clearer accountability for their decisions as those systems gain the ability to communicate and act independently.
Legacy systems and data remain critical
Several speakers stressed that AI transformation depends on the quality of existing enterprise systems and information. The issue received detailed attention in the “Modernizing the Enterprise Application Estate” session, moderated by Woody Evans, vice president of global solutions engineering at Perforce Delphix.
Ron Guerrier, chief technology officer of Save the Children, warned that AI-generated applications can create substantial operational risks when business teams expect prototypes to move directly into production. He advocated established engineering practices and continuous improvement principles, along with investment in employees who understand the organization’s technology environment and business processes.
Deepak Gangwaar, senior vice president at Northern Trust, highlighted the importance of reliable, accessible test data and appropriate compliance controls. Faster code generation offers limited benefits if developers must still wait for suitable data to validate applications, he noted.
Panelists also explored AI’s potential to help organizations analyze legacy code, understand embedded business logic and identify modernization opportunities.
Guerrier cautioned that enterprises must weigh the full economics of modernization, including ongoing infrastructure expenses, AI consumption costs and long-term vendor commitments.
CIOs reconsider how they invest
Uncertainty in AI technology markets prompted a discussion of whether CIOs should adopt some of the investment practices venture capital firms use.
Mike Asem, founding partner at M25, and Gale Wilkinson, founder and managing partner of Vitalize, said organizations can benefit from testing promising technologies through structured pilot programs with defined objectives and acceptable levels of risk.
Asem pointed to the opportunity cost of avoiding emerging technologies and the value of validating assumptions before committing substantial resources.
Wilkinson described growing investor interest in specialized AI applications that address specific business processes. She suggested that enterprises can gain by working with early-stage technology companies as design partners, helping shape solutions around operational requirements.
Their remarks highlighted the need to balance governance and risk management against the possibility that excessive caution could leave organizations behind competitors.
Leadership and workforce development take center stage
The conference closed with a discussion of how CIO responsibilities may evolve over the next 12 to 24 months.
Samir Shah, senior vice president and global CIO of Brunswick Corp., described an AI operating model in which business executives identify priority processes, commit funding and take responsibility for financial outcomes.
Shah said Brunswick has generated about $8 million to $10 million in tangible savings over the past 18 months. The company is pursuing a portfolio of 18 to 20 AI use cases across areas including supply chain operations, customer service, pricing and software development, with more than $50 million in targeted value.
Darlene Reina, director of IT and AI enablement at Babylist, said her organization is giving business employees the tools and guardrails to build applications themselves, freeing IT specialists to concentrate on higher-value initiatives and production requirements.
Reina also addressed the workforce implications of AI adoption. As routine technology tasks become automated, she said, organizations must prepare existing employees for new responsibilities and rethink how they recruit and develop entry-level talent.
She advocated retraining and redeploying employees whose responsibilities are changing, noting that their knowledge of business operations remains valuable as technology capabilities evolve.
Speakers throughout the day identified financial discipline, governance, sound technical foundations and workforce development as the requirements for sustaining AI-driven transformation as the technology continues to evolve.