The Executive’s Guide to Agentic AI: Enhancing Workflows to Unlock High-Value Potential

Over the last few years, implementing enterprise AI has felt like building a skyscraper while the blueprints are continuously being redrawn. The rapid rise of ChatGPT in early 2023 thrust generative AI into the spotlight, prompting businesses to reconsider their digital strategies. As we moved through 2024, the initial hype gave way to a more pragmatic approach, with organizations seeking tangible ways to safely and securely integrate AI into their operations. Now, as we venture into 2025, a new technology is gaining traction in enterprise AI, one that feels a step closer to the early, thrilling promise of artificial intelligence — AI with agency.

This agentic AI is capable of autonomous decision-making and complex task execution with minimal human oversight. In our recent webinar, The Executive’s AI Playbook: Lessons from 2024 & Strategies for 2025, our panel of experts, including AI thought leader and influencer Allie K. Miller, agreed that 2025 will be the year agentic AI moves from concept to practical application. This shift promises to reshape how enterprises use workflows across all sectors.

Understanding this technology will be crucial to staying competitive in the evolving digital landscape. So, let’s get to know agentic AI a little better and understand how you can use AI agents to transform your operations and unlock new value.

The Growing Need for Tangible AI Returns

As AI technology advances, enterprise investments are keeping pace. Take a look at these projections across three key industries:

• Financial Services: Worldwide investment in AI is expected to more than double by 2027. (Statista)
• Healthcare: AI investment growth is forecasted at an annual rate of nearly 50% over the next five years. (Morningstar)
• Telecommunications: AI investments are predicted to grow from $3.7B to $9.3B in the next four years. (Horizon Databook)

AI is no longer viewed as a novelty. Executives see the technology as a powerful force in accomplishing work, and their substantial investment brings a pressing need for clear key performance indicators (KPIs) and measurable return on investment (ROI) in 2025.

AI agents will be pivotal in unlocking measurable advantages through cost reductions and revenue growth. That’s because they enable enterprises to move beyond individual employees using generative AI for isolated tasks. Instead, entire teams can leverage AI for complete workflow management. While that potential for significant increases in revenues and profits is undoubtedly attractive, the technology’s true promise lies in its ability to free up professionals to focus on high-value, specialized work that requires uniquely human expertise.

What Makes Agentic AI Different from Generative AI?

If the arrival of AI agents represents another hurdle in incorporating AI into your organization, it may be helpful to think of it as a natural progression. Agentic AI, referred to as the third wave of AI, builds upon the foundations laid by predictive AI (first wave) and generative AI (second wave).

Most people are now familiar with generative AI as a tool — prompted by human operators — that excels at creating content, like text, images, or code. However, its capabilities are limited to producing single, discrete outputs by mimicking patterns from its training data. While it’s powerful in many respects, generative AI cannot autonomously complete complex objectives or workflows.

2024 saw the emergence of Retrieval Augmented Generation (RAG), a significant advancement that enhanced the capabilities of LLMs. RAG allows AI to access and summarize relevant, up-to-date information beyond its initial training data. This development improved the accuracy of AI outputs and enhanced more sophisticated applications, such as AI copilots. An everyday use case for copilots is in coding, where a human programmer receives notifications in real-time with suggestions to improve the code or fix potential bugs and issues.

Unlike copilots that support human-driven tasks, AI agents can “take the lead,” independently planning and executing various functions to complete work. Key characteristics of an AI agent include:

• Autonomy: Creating and executing an action plan once initiated.
• Process Automation: Managing a sequence of tasks, not just single outputs.
• Self-revision: Iterating and improving its own outputs.
• Objectives: Working towards broader goals rather than responding to specific prompts.

With AI agents managing workflows, complex business processes can be streamlined, and bottlenecks resolved. And as agents work semi-autonomously, businesses can scale operations more effectively.

Applying Agentic AI to Workflows in Three Industry Use Cases

Understanding agentic AI’s potential becomes clearer through real-world applications. Let’s examine how this technology can address specific challenges in Financial Services, Telecommunications, and Healthcare — transforming workflows in ways that traditional generative AI cannot achieve alone.

Financial Services

• Sample pain point: Generating required documentation (both internal and external) is a time-consuming task for a banker meeting with an enterprise client who is requesting a commercial loan.
• Generative AI solution (Current): A lender or underwriter uses discreet tools for customer research, data synthesis, and visualization, each requiring individual review and compilation.
• Agentic AI solution (Future): An AI sales agent proactively searches internal customer relationship management (CRM) databases and lending platforms, requests financial data, investigates social media for reputational posts about the client, and generates an initial credit recommendation.

Telecommunications

• Sample pain point: Average revenue per user (ARPU) decreases due to customer perception that a telecommunication service provider’s (TSP) offering is a commodified utility.
• Generative AI solution (Current): A customer experience (CX) agent manages customer interactions by using suggestions from generative AI about tone and sentiment to offer a solution from an FAQ document.
• Agentic AI solution (Future): A human CX agent engages an agentic AI interface to offer personalized upsell/ cross-sell offers that are contextualized on a customer’s profile with usage patterns and call support intent.

Healthcare (Payer)

• Sample pain point: Claims adjudication is manual, complex, subject to abuse, and can lack accuracy.
• Generative AI solution (Current): A claims adjudicator leverages AI to identify high-risk providers by analyzing patterns in claim frequency and fraud severity, creating prioritized investigation lists.
• Agentic AI solution (Future): An AI claims adjudicator agent identifies miscodes, automatically notifies providers of correction requirements, and schedules follow-up reviews after two weeks.

Unlocking High-Value Work

In each case above, agentic AI gives humans more time to pursue high-value tasks, activities, and responsibilities that directly contribute to an organization’s strategic goals. High-value work often emphasizes uniquely human capabilities such as empathy, intuition, and cross-contextual thinking. It can involve navigating ambiguity and making nuanced decisions considering quantitative data and qualitative factors.

In our industry examples, agentic AI might unlock:

Opportunities for bankers and portfolio managers to:

• Improve relationships with internal stakeholders and external agencies by streamlining compliance and regulatory reporting.
• Conduct more frequent portfolio reviews and manage risk by proactively analyzing large volumes of legal and underwriting documents.
• Analyze customer complaints and provide reporting to compliance and audit groups to flag potential “matters requiring attention” before regulatory agencies get involved.

Efficiencies that help departments within a TSP to:

• Focus on proactive customer experience enhancement like designing VIP programs and implementing strategic initiatives that boost satisfaction and loyalty.
• Dedicate more time to strategic infrastructure planning, analyzing technology trends, and future-proofing network architectures that maintain a competitive advantage.
• Build service packages that combine mobile, internet, and smart home solutions to meet emerging customer needs and increase lifetime value.

Bridge-building across healthcare partnerships, fueled by:

• Data-driven policy changes that reduce claim denials, improve patient satisfaction, and free providers to offer better care.
• Long-term financial strategies for healthcare organizations that use more accurate and efficient claims data.
• Education initiatives that build trust with patients by helping them better understand their insurance coverage and how to navigate the healthcare system.

By focusing on high-value work, organizations can maximize time for initiatives driven by human expertise and creativity. This drives innovation and helps build sustainable competitive advantages.

Building Blocks of an Effective AI Agent

With a clear understanding of agentic AI’s potential, let’s explore how to create effective agents for your enterprise. While development tools are readily available, successful implementation requires understanding three essential building blocks in creating an AI agent. Identifying these foundational elements will help you make informed strategic decisions before beginning development.

1. Purpose and goals

At its core, an AI agent must have a clearly defined mission that guides its decisions. This purpose establishes what the agent should accomplish, how it should prioritize tasks, and how success will be measured.

Financial Services example: An agent supporting investment analysts at a brokerage firm could be tasked with continuously monitoring global financial news, generating daily briefings of market-moving stories, and assessing their potential impact on client portfolios. Its primary goal would be to enhance decision-making by delivering timely, relevant insights and data-driven projections for short-term market trends, enabling analysts to anticipate market shifts more effectively.

2. Knowledge base

Every AI agent needs access to a carefully curated collection of information sources and databases that inform decision-making and task execution. This knowledge base might include industry-specific data, regulatory requirements, internal policies, and info feeds essential for the agent’s designated role.

Healthcare example: An agent designed to assist physicians in medical diagnosis would require access to a comprehensive medical database, which includes symptoms, diseases, treatment protocols, and drug interactions. It would also need up-to-date patient records, recent medical research findings, and relevant healthcare regulations.

3. Tooling

An AI agent must be able to utilize the tools needed to accomplish its goals. It may be charged with calling and opening various programs and apps (including generative AI), completing a host of forms that all present distinct challenges, and synthesizing information from documents and websites that vary widely in style and format. This may be the most challenging aspect of agentic AI development because a practical interface between an agent and other digital tools must be designed.

Telecommunications example: An AI agent designed to help human CX agents uses RAG connections to CRMs, historical call and chat transcripts via customer support software, real-time product and pricing tables, FAQs, and other knowledge bases. With access to all relevant tools, the AI agent builds complete customer profiles that include upsell and cross-sell opportunities and personalized sales scripts for CX agents to engage with customers in more helpful ways.

The adaptability of AI agents stands out as a key advantage for enterprises. Some agents are designed as out-of-the-box solutions for standard functions like customer service or regulatory compliance, while others are customized for industry-specific applications. Across all applications or industries, the effectiveness of AI agents hinges on their context-awareness — their ability to make informed decisions based on predefined goals, access to essential tools, and the latest data from diverse sources.

Preparing with “Think Slow, Act Fast” and “Human in the Loop” Strategies

As the technology evolves across 2025, enterprise adoption of agentic AI will surge. However, this rush mentality carries inherent risks, particularly for security and governance. In industries where high-risk data is involved, a “think slow, act fast” strategy is advisable. Organizations should carefully evaluate and select the right tools for their specific needs, then move decisively with implementation. Regardless of data sensitivity levels, successful deployment requires thoughtful integration of human oversight — the human-in-the-loop (HTL) approach.

Agentic AI will be hampered by bugs, challenges, and limitations in its infancy, but its capability and reliability will likely follow a similar path to generative AI, becoming more enterprise-secure over time. However, to effectively integrate AI with humans-in-the-loop, organizations will need to prepare now. Even with the promise of some level of autonomy, agents will still need to be properly deployed, calibrated, and trained throughout their lifecycle. Just as a generative AI tool needs a qualified individual to properly use it, AI agents require the right “guides” within an organization to ensure value is realized.

Finding the “Sweet Spot” for Implementation

Given the broad capabilities of AI agents, the real question for business leaders is where to apply this technology. Identifying the best uses for agents will require the evaluation of two key factors — value and trust.

Value

Agentic AI has instant appeal when evaluating the bottom line, and it’s not hard to imagine the impact agents could have on revenue growth (sales, marketing, and customer concierge) or cost reductions (customer service, fraud monitoring, and transaction processing). The value of AI agents will be as a force multiplier, opening up opportunities to scale and streamline existing operations without adding significant overhead.

Trust

In highly regulated sectors such as Healthcare, Telecommunications, and Financial Services, where handling sensitive personal data is integral to operations, trust becomes as crucial as value. The potential applications for AI agents are numerous, but success depends on cultivating and maintaining trust across an entire ecosystem. It’s not just about data security; this level of trust encompasses transparency in AI decision-making, adherence to ethical standards, and the ability to demonstrate consistent, reliable performance.

To identify the sweet spot for your use of agentic AI, evaluate opportunities for revenue growth or cost reductions, and then narrow down those use cases where trust is maintained or enhanced using AI. What are the blockers that prevent talented professionals within your organization from building more meaningful relationships and pursuing innovation? This may reveal the areas where AI agents could free teams up to spend more time doing what they do best.

While agents optimize the back-office, bottom-line value functions, team members can spend their time at the “front of the house,” building a level of trust that’s a differentiator in a world where AI agents become the norm.

From Big Picture to Brass Tacks

Big picture: Agentic AI is not just about automating routine tasks; it’s about fundamentally improving how businesses operate, make decisions, and create value. The challenge for companies will be to identify and engage solutions that enhance the quality of human work on top of driving efficiency and profitability. Those that do will be well-positioned to emerge as leaders in the new era of AI-enabled business.

Brass tacks: The journey from potential to measurable success requires careful planning and evaluation. In our next article on agentic AI, we’ll share WillowTree’s practical formula for evaluating the tangible returns of implementing AI agents. Continuing our deep dive into Financial services, Telecommunications, and Healthcare, we’ll provide a framework for calculating ROI in these essential industries.

Unlock the potential of agentic AI in your industry. Whether you're just starting or ready to optimize your current AI strategy, our Data & AI Consulting team is here to help. Contact WillowTree today.