In a recent Forrester survey of more than 16,000 purchase influencers, we found that 14 percent say that artificial intelligence investments are not a priority. That data point is unsettling because organizations that don't urgently move to adopt AI won't be able to capitalize on the wave of agentic AI that is poised to transform customer operations. It means that these late adopters won't be able to easily pivot their businesses to align with changing customer expectations and economic conditions, and they'll risk losing market relevance over time.
There are two categories of AI: Predictive AI analyzes existing data to make forecasts (think segmentation, best offer, or next-best step) and has been around for decades. Generative AI (genAI) is recent, and it creates content based on learned patterns (think service replies, case history summary). These are the foundational components of an AI strategy.
AI agents are the rage now, but they don't fundamentally introduce a new type of AI. They use both predictive and generative AI in combination with rules to pursue business goals autonomously. AI agents also learn from outcomes and adapt their behavior over time.
Vendors are rolling out AI agents at break-neck speed. They are also offering tools to create custom agents and to orchestrate agents working together. But how do you start your journey to AI agent operations?
Step 1: Start small by automating simple rules-based tasks. Look for opportunities to deploy AI agents that can handle simple knowledge retrieval tasks or follow defined workflows and recommend next actions like an interaction with a human agent or logging a ticket.
Step 2. Let AI agents execute tasks with well-defined guardrails. Pinpoint more complex workflows that might already be supported by several discrete AI agents. For example, an AI agent that combs through submitted resumes, evaluates resumes against job description criteria, and sends the best ones to a manager. Or an AI agent that finds similar tickets and recommends troubleshooting steps to a human agent. These agents have built-in learning loops.
Step 3. Expand the remit of AI agents to broader workflows. Look for cross-departmental workflows that synthesize data from different sources. For example triggering customer outreach when triangulating on new product opportunities within an account, customer health, ticket volume, and financial data such as contract details and payments.
This, of course, takes a solid foundation of data management, process definition, and management, governance, and change management to get your human agents ready to collaborate with AI agents. It's no easy journey!
Kate Leggett is a vice president and principal analyst at Forrester Research.