We may look back at 2024 as the year when LLMs became mainstream, every enterprise SaaS added copilot or virtual assistant capabilities, and many organizations got their first taste of agentic AI. CIOs were given significant budgets to improve productivity, cost savings, and competitive advantages with gen AI. AI at Wharton reports enterprises increased their gen AI investments in 2024 by 2.3 times compared to 2023 but forecasts lower increases over the next two to five years.  

Experienced CIOs know there is never a blank check for transformation and innovation investments, and they expect more pressure in 2025 to deliver business value from gen AI investments. Deloitte’s State of Generative AI in the Enterprise reports nearly 70% have moved 30% or fewer of their gen AI experiments into production, and 41% of organizations have struggled to define and measure the impacts of their gen AI efforts. As gen AI heads to Gartner’s trough of disillusionment, CIOs should consider how to realign their 2025 strategies and roadmaps.

CIOs feeling the pressure will likely seek more pragmatic AI applications, platform simplifications, and risk management practices that have short-term benefits while becoming force multipliers to longer-term financial returns. CIOs should consider placing these five AI bets in 2025.

Scale trusted workflows with agentic AI

Appian, Atlassian, Cisco Collaboration, Forethought, IBM, Ivanti, Pega, Salesforce, SAP, ServiceNow, Tray.ai, Workday, Zoho, and others launched service-oriented AI agents in 2024. Whereas robotic process automation (RPA) aims to automate tasks and improve process orchestration, AI agents backed by the company’s proprietary data may rewire workflows, scale operations, and improve contextually specific decision-making.  

While agentic AI promises efficiency and scalability, many question whether autonomous agents are enterprise-ready. The World Economic Forum shares some risks with AI agents, including improving transparency, establishing ethical guidelines, prioritizing data governance, improving security, and increasing education. 

Organizations can opt for a pragmatic approach and seek human-in-the-middle AI agents that work collaboratively with decision-makers and subject matter experts. A human-centric approach helps with the change management efforts around using agentic AI while evaluating the benefits and risks. 

“In 2025, companies at the forefront of the agentic AI revolution will face a critical challenge: balancing the delivery of seamless, done-for-you experiences with the need to give customers ultimate authority and control over final decision-making at their discretion,” says Ashok Srivastava, chief data officer at Intuit. “To achieve this, innovation must focus on AI systems that seamlessly blend advanced autonomy with user-centric control, incorporating adaptive transparency, ethical safeguards, and context-aware learning to empower customer decision-making.”

What to bet on: Expect significant agentic AI hype in 2025 on one end and potential employee fears around autonomous agents taking their jobs on the other. CIO should bet on change management programs and evangelizing high-quality agents with whom employees collaborate to deliver value beyond productivity.

Build toward intelligent document management

Most enterprises have document management systems to extract information from PDFs, word processing files, and scanned paper documents, where document structure and the required information aren’t complex. Examples include scanning invoices, extracting basic contract information, or capturing information from PDF forms. Even simple use cases had exceptions requiring business process outsourcing (BPO) or internal data processing teams to manage. 

Michael Beckley, CTO and founder of Appian, says document processing is a boring gen AI use case with significant business potential. “With traditional OCR and AI models, you might get 60% straight-through processing, 70% if you’re lucky, but now generative AI solves all of the edge cases, and your processing rates go up to 99%,” Beckley says.

Many legal departments can benefit from intelligent document management where the time to review contracts impacts operations. Even simple contracts like non-disclosure agreements can require 30-day turnarounds, according to a Bloomberg Law study.

“Legal AI enhances compliance and speeds up traditionally lengthy processes, such as contract negotiations and monitoring for contract renewals, enabling legal teams to work more efficiently and ultimately increase revenue,” says Anurag Malik, president and CTO of ContractPodAI. “AI-driven tools streamline workflows and reveal valuable insights, allowing organizations to manage contract reviews, risk analysis, and compliance with greater efficiency.”

Other document processing use cases include conducting clinical trials in life sciences, loan underwriting in retail banking, and insurance claims processing. Workflows with these documents require deep subject matter expertise that can be aided or accelerated with gen AI document processing capabilities.

Vinay Samuel, CEO and founder of Zetaris, shared with me one intriguing integration of agentic AI with document processing that can have profound corporate governance impacts. “With agentic AI, a board can have a virtual board member that can access, analyze, and interpret thousands of documents in real-time, which may result in a golden age of compliance and corporate governance,” Samuel says.

What to bet on: Look for scalable departmental opportunities with complex business rules embedded in document processing and a mix of no-code, low-code, RPA, and BPO solutions in place. Reengineering these workflows with ground-floor gen AI capabilities can deliver cost benefits and also help the IT department consolidate platforms. 

Prioritize marketing’s customer data needs

CIOs looking for growth opportunities from gen AI investments should start by reviewing the marketing department’s objectives and integration challenges. According to The State of Martech 2024, companies with over 10,000 employees average 650 SaaS applications, while smaller enterprises average just under 300. The report shows portfolio consolidation and integration investments over the past year, yet only 32% claim that over 80% of their marketing stack is integrated.

Why focus on the marketing department? While many gen AI efforts focus on productivity and employee experiences, marketing departments should have examples where gen AI investments yield customer experience improvements and revenue growth.

One opportunity is for CIOs to help their marketing departments improve brand loyalty. Mike Lee, president and GM at AND Digital, says, “In the travel and loyalty industry, generative AI is revolutionizing how customers interact with reward programs.” Lee described an AI travel agent driving increased bookings with an intuitive AI-powered product that 83% of users preferred over traditional search options, propelling daily profits above $1 million.

Jaime Meritt, chief product officer at Verint, shares a second example of using AI in customer contact centers to save millions of dollars through CX automation, driving efficiencies, consistency, accuracy, and compliance. “Brands that delay adopting and deploying AI that delivers business outcomes in their contact centers will start to experience lower CX scores and attrition, which can create competitive lag and dramatic financial impact,” Meritt says.

Placing an AI bet on marketing is often a force multiplier as it can drive data governance and security investments. “AI innovation can not — and should not — exist without parallel investment in governance to ensure its responsible and effective integration,” says Henry Umney, MD of GRC strategy at Mitratech.

What to bet on: Marketing’s data and workflow integrations are an abyss of technical debt because of having too many SaaS tools and the ease of running experiments. CIOs should prioritize objectives tied to measurable improvements in customer experience and accelerated sales outcomes, then look for opportunities where winning AI capabilities can drive stakeholder consensus on platform consolidation. 

Realign from being data-driven to AI-driven

Enterprises often have centralized data science teams for developing AI models, reporting teams aligned with the various enterprise platforms, citizen data scientists trained on data visualization, dataops departments overseeing data pipelines, and proactive data governance functions to develop policies and ensure compliance. Even this breakdown leaves out data management, engineering, and security functions.

That’s a lot of moving organizational parts, and CIOs may seek to use gen AI as a driver behind a cohesive strategy, organizational model, and platform capabilities, especially when seeking industry-specific AI and analytics differentiating capabilities. “By embedding AI into data analysis frameworks, organizations can unlock unprecedented capabilities in healthcare diagnostics, manufacturing quality control, and marketing optimization, turning raw data into strategic competitive advantages,” says Ashwin Rajeeva, co-founder and CTO of Acceldata.

Why should CIOs bet on unifying their data and AI practices?

In 2024, departments and teams experimented with gen AI tools tied to their workflows and operating metrics. It created fragmented practices in the interest of experimentation, rapid learning, and widespread adoption — and it paid back productivity dividends in many areas. However, data-driven organizations can use 2025 as a year to realign their data, analytics, and AI efforts to seek out more strategic benefits.  

“You can imagine that in 2025 and beyond, data work can be delegated entirely to AI but checked or corrected by human colleagues, or data experts can rely on AI to give intelligent feedback based on the collective wisdom of data teams across the enterprise,” says Michael Berthold, CEO of KNIME. “However, this is only possible if you invest in technology that brings transparency and reliability to AI-performed or AI-assisted data work.”

What to bet on: CIOs should look for operational and risk management gaps in their AI and data strategies and consider bets on data fabrics, data observability, analytics platforms, AI governance, and data security posture management.

Revisit the IT stack to support AI objectives

Marketing departments aren’t the only ones where having too many SaaS tools and using spreadsheets for integrations can become a barrier to adopting AI solutions. “CIOs are likely to find a glut of digital tools that create information silos and impact strategic decision-making, says Dalan Winbush, CIO of Quickbase. “Disconnected data creates unnecessary gray work as employees hop from app to app searching for the information they need to do their job.”

IT and devops teams suffer similar tool proliferation that may have been acceptable in the devops glory years, where many development teams selected their tools with few constraints. Many organizations are shifting to platform engineering to improve developer experience and productivity.

A third area of shifting debate concerns infrastructure between data centers, public clouds, hybrid clouds, multicloud, and edge computing. Over the past decade, CIOs and CISOs shifted their strategies based on ease of use, scalability, security, and costs, only to find that their golden rules for selecting optimal architectures yielded many exceptions and evolved yearly with infrastructure innovations.  

So where should CIOs place their infrastructure, application rationalization, and data centralization bets to better support this next era of agentic AI? Should CIOs bring AI to the data or bring data to the AI?  

What to bet on: Infrastructure choices and optimized architectures depend on the use case, but several disciplines stand out for CIOs to bet on in 2025 to shape their technology strategies and plans.

• Having too many tools, manual processes instead of integrations, and significant technical debt inhibit where and how well AI can be applied effectively. CIOs facing these challenges should consider consolidation and platform strategies.
• Before gen AI, speed to market drove many application architecture decisions. In the AI era, it may be speed to accessible, secure, and high-quality data that drives data management strategies.
• Given the number of infrastructure, platform, and operational choices, CIOs will need a strong finops practice to model costs, especially as AI capabilities scale data and processing needs.

Given how fast AI capabilities are evolving, I expect more CIOs will adopt a “less is more” approach in 2025 to selecting and managing technologies and bet on platforms offering greater interoperability and portability.