January 26, 2026 • 10 Min Read

Gartner forecasts that by 2027, over 40% of agentic AI initiatives will be abandoned due to unclear objectives and weak execution (Reuters, 2025). That failure rate isn’t about the technology; it’s about how it’s applied.

CFOs, however, remain bullish on AI’s promise. In RGP’s latest survey, 66% of CFOs expect measurable AI impact within two years, yet only 14% are seeing meaningful ROI today. That disconnect is telling. The biggest barrier isn’t innovation—it’s alignment. Many AI efforts fall short because they’re treated as tech deployments, not as fundamental shifts in business design. They automate low-value tasks, rely on poor data, and neglect the human workflows they aim to enhance.

CFOs are uniquely positioned to bridge this gap. They control the levers of capital, risk, and performance measurement, They are at the heart of what AI most urgently needs: clarity of purpose and confidence in data. Their role is no longer just to fund or evaluate AI; it’s to translate ambition into accountable, scalable results.

Why AI Pilots Fail (and What CFOs Can Learn From It)

Most failed AI initiatives don’t fail because of technical flaws—they fail from not understanding the real problems to be solved. Common mistakes include:

• Automating low-value work: Many AI pilots chase efficiency instead of solving business-critical problems such as growing your customer base, optimizing pricing, and ensuring your leaders have the real-time insights they need to make smarter decisions.
• Relying on untrusted data: With only 19% of CFOs fully trusting their data (RGP’s 2025 CFO Perspectives), it’s clear: before AI can deliver real value, organizations must rebuild trust in their data. Solve the data problem now, and AI becomes the natural next step.
• Implementing without user input: Tools that overlook how people truly work and the daily challenges they face risk doing the opposite of what they’re meant to — creating more work instead of less.

This pattern isn’t unique to AI—it mirrors the same root causes behind failed automation and transformation initiatives.

What Works: A Human-Centered Path to ROI

The AI programs that succeed share one defining trait: they’re designed around people. From the leaders who need reliable, real-time insights to make better decisions to the employees struggling to navigate new systems, disparate resources, and time-consuming tasks, all problems worth solving lead back to the people experiencing them daily.

Can AI help solve all of these? Absolutely. The challenge is getting there, and human-centered problems require human-centered solutions. By understanding what the people in your organization need to do their jobs not just effectively but delightfully, not just automated but empowered, you will know how to apply AI successfully and see a return on your investment.

Your Framework for AI Pilot Success

Human-centered design (HCD) principles—long the foundation of great customer experiences—are now the key differentiator in enterprise transformation. As organizations explore AI pilots, success depends on more than the technology itself. Embedding HCD ensures solutions align with how people actually work, driving adoption, effectiveness, and sustainable ROI.

HCD Principles

Begin by deeply understanding the people who will use or be impacted by AI to identify where it can support—not replace—human decision-making. This will also define your success metrics from both user and business viewpoints.

Achieve this by conducting interviews, shadowing daily activities, and mapping out the real, not expected, process and points of friction. Remember to include diverse perspectives across departments and employee types to prevent bias and ensure fair outcomes.

Once you’ve identified the problems AI could solve, the next step is deciding which ones it should solve. The real question becomes: which challenges are truly holding your organization back from achieving its strategic goals?

If your company is prioritizing revenue growth, look to the friction points impacting your clients’ and sales teams’ experiences. If cost reduction is the goal, explore where AI can go beyond automating manual work to actually eliminating it—through self-service capabilities for customers or employees.

Not every pain point is worth automating or enhancing with AI. Teams should evaluate potential use cases based on their alignment with strategic objectives, measurable business value, and real user needs. By prioritizing efforts that will move the needle most—for both the business and the people it serves—organizations can focus resources where AI will create the greatest impact and build early momentum for adoption.

Once you have your sights set on the right problem to solve and you’re ready to design the solution, prototype the experience before the algorithm. Visualize how AI fits into daily workflows with simple mockups or simulated outputs before committing time or money. Are employees having to go to multiple systems to pull information to complete a task, which an AI could do for them? Or are they spending time manually creating content such as sales materials or training guides, which an AI trained on your business could accelerate?

Imagining these scenarios with employees, domain experts, and stakeholders early and often through design feedback sessions can quickly target the capabilities that will provide the most value. You will likely also discover what users need to build trust in how the AI generates insights or recommendations.

During development and preparing for deployment, keep your users close. Test throughout the development process to refine both the model and user experience continuously based on real-world use. Remember, too, to gather feedback on usability, trust, and value—not just accuracy. Just because it works doesn’t mean it’s useful or achieving the impact promised. Getting feedback early and often from the people who will use it will allow you to make the right adjustments and create a change management plan to ensure success and adoption.

How to Get Started

Every organization needs a practical framework for evaluating every AI investment. The difference between AI that scales and AI that stalls comes down to alignment and preparation—of design, data, and people.

Ask yourself the following questions when planning an AI pilot to ensure you can get the expected value:

Strategic Fit

• Key question: Does this AI solve a high-value business problem—or chase novelty?
• What to do: Run an AI value-mapping session with business and tech leaders.
• Short-term value (0–12 months): Early visibility into ROI-driven use cases.
• Long-term value (3–5 years): Continuous prioritization of evolving opportunities.
• How RGP can help: Strategy & Transformation — Identify and prioritize ROI-driven use cases.

Reliable Data

• Key question: Can we trust our data sources?
• What to do: Conduct a data readiness and governance audit.
• Short-term value (0–12 months): Immediate data quality improvements.
• Long-term value (3–5 years): Foundation for scalable AI and advanced analytics.
• How RGP can help: Data & Analytics — Build governance frameworks for AI reliability.

Human Experience

• Key question: How will this affect employee workflows and decision-making?
• What to do: Use human-centered design (HCD) workshops to co-create AI-enabled processes.
• Short-term value (0–12 months): Improved user adoption and reduced change resistance.
• Long-term value (3–5 years): Sustainable workforce trust and operational agility.
• How RGP can help: Human-Centered Design (HCD) — Ensure intuitive, inclusive adoption.

Measurement

• Key question: How will we measure success beyond launch?
• What to do: Define a value realization framework with KPIs for adoption and efficiency.
• Short-term value (0–12 months): Early ROI indicators and efficiency baselines.
• Long-term value (3–5 years): Ongoing ROI tracking tied to business outcomes.
• How RGP can help: Finance & Shared Services Optimization — Build dashboards for continuous ROI tracking.

Risk & Governance

• Key question: Are we deploying AI responsibly?
• What to do: Integrate responsible AI governance frameworks (e.g., NIST, ISO/IEC 42001).
• Short-term value (0–12 months): Immediate regulatory compliance and risk mitigation.
• Long-term value (3–5 years): Long-term stakeholder trust and brand resilience.
• How RGP can help: Governance, Risk & Compliance (GRC) — Embed trust and compliance into AI.

The CFO as Architect of Value

The secret behind AI maturity is shifting from how to why. Instead of focusing on how to get up and running with AI, think about why you need AI in the first place.

CFOs are not just investors in AI and other technologies—they are architects of the value these systems are expected to provide, which should balance today’s return on capital with tomorrow’s return on capability.

The future of finance leadership won’t be defined by who adopts AI fastest—but by who applies it most wisely.

• Short Term: Align AI with measurable financial outcomes—efficiency, accuracy, and productivity.
• Long Term: Invest in the enablers of future-ready performance—trusted data, ethical frameworks, and human-centered design.

CFOs who treat AI as both a strategic investment and a human-centered transformation unlock its full potential: reliable data, empowered employees, and measurable impact.

When designed responsibly, AI amplifies what humans do best—judgment, creativity, and empathy. The most successful CFOs aren’t modernizing systems; they’re designing organizations that are smarter, more adaptable, and deeply human.