Using AI Agents to Create a Culture of Adaptability
AI agents, sophisticated systems trained on organizational data, are transforming businesses. Unlike simpler AI assistants, agents handle complex tasks autonomously, learning and adapting from interactions to improve efficiency. This post explores various types of AI agents, from basic chatbots to complex hierarchical systems, highlighting their applications across numerous industries. While acknowledging implementation challenges, we emphasize the significant benefits of AI agents in boosting productivity, streamlining processes, and enhancing customer experiences, ultimately arguing that the potential gains outweigh the risks.
Listen to the “Using AI Agents to Create a Culture of Adaptability” Podcast
In the latter half of 2024, discussions and predictions for 2025 increasingly focused on the emergence of “AI agents.” These agents are being deployed to train customer service representatives, educate employees on internal policies and drive online sales. Salesforce reported that during Cyber Week, when orders were up by almost 8% year over year, AI, including agents, “played a pivotal role, influencing $60 billion in sales and showing how this technology is reshaping the retail landscape and boosting consumer engagement.” The same report cited that “retailers using generative AI and agents to fuel service experiences saw a 2% higher conversion rate compared to retailers who didn’t use the technology.”
“AI, including agents, “played a pivotal role, influencing $60 billion in sales and showing how this technology is reshaping the retail landscape and boosting consumer engagement.”
Several professional services companies are using AI agents to simplify time-consuming internal processes, such as retrieving company policies, speeding up the hiring process by pre-qualifying candidates, creating marketing materials, onboarding new employees and providing self-help customer service.
AI Agents vs. AI Assistants
AI agents are sophisticated systems trained on an organization’s proprietary data and domain expertise. They are designed to perform tasks ranging from simple prompt-response interactions to complex end-to-end business workflows. As industry experts note, “Agents can think or reason, remember, be trained, and even know when to ask for help.” These agents can be programmed to send emails, create support tickets, update records and respond to specific event triggers or customer requests. Importantly, they are trained to recognize when they need further clarification or guidance.
The most autonomous AI agents tackle more complex functions. They can independently choose a course of action, design a plan, gather relevant data and employ multiple software tools to complete each step.
By contrast, AI assistants, like ChatGPT, Perplexity or Microsoft’s Copilot, are designed to handle complex queries, engage in conversational interactions and perform a wide range of tasks. Built on advanced natural language processing, these assistants are engineered to learn and adapt from user interactions over time.
AI agents are designed to continuously learn and adapt based on previous interactions and to apply this learning to future interactions. This allows them to remain proactive and relevant despite changes in consumer or business preferences. The more data fed into the system, such as customer conversations, transaction histories and social media interactions, the more effective the agent becomes in addressing customer queries. This ongoing influx of data not only teaches the agent to identify patterns and make the right decisions, but also enables it to learn from previous responses. These capabilities allow the agent to provide timely and accurate answers, or to determine if the customer needs to be escalated to a human representative.
Platforms like Meta AI now allow users to select from various voice options for assistants, including well-known voices, such as Awkwafina, Dame Judi Dench and John Cena. Similarly, Eleven Labs enables users to listen to their favorite books or articles narrated by iconic voices—for example, Judy Garland, James Dean and Sir Laurence Olivier—offering hundreds of voices in 32 languages.
Taking this a step further for the same healthcare client, I recently explored Google’s NotebookLM platform, which can generate AI-hosted podcasts. Featuring male and female hosts engaging in a realistic dialogue, this platform uses uploaded content to create an immersive podcast experience.
If you’ve already listened to the AI-generated audio at the top of this post, created using NotebookLM, you’ll find the results both impressive and somewhat unsettling in their authenticity.
Which AI Agent Is Best for You?
Source: Amazon AWS
There are many types of AI Agents that can be implemented, ranging from the most basic, like a customer chatbot, to more sophisticated agents that learn and improve over time. Below, are a few and their potential applications:
Simple Reflex Agents:
These agents are built on condition-action rules that are limited by the subject matter—for example, customer service chatbots that provide basic responses to common queries in financial institutions or retail.
Model-Based Reflex Agents:
These agents use a model of their environment to inform decisions due to their ability to think through potential outcomes and barriers by utilizing reactive decision-making with contextual awareness—for example, Smart thermostats that can harness current temperature readings and time to predict future settings for heating or cooling.
Utility-Based Agents:
These agents evaluate multiple potential actions and choose the one that offers the “best” outcome based on a utility function—for example, a driving route recommendation system that provides the “best” route.
Goal-Based Agents:
These agents are designed to achieve specific goals and outcomes. They are forward-thinking in the decision-making process to determine their actionable goals—for example, in healthcare, goal-based agents can improve patient outcomes by assisting in diagnosis, treatment and patient monitoring.
Learning Agents:
These agents continuously learn over time while also adapting to changing conditions to improve performance—for example, Netflix and Amazon use our viewing habits, ratings and searches to serve up viewing or product recommendations.
Hierarchical Agents:
These are essentially a group of AI agents organized by tiers or levels, where the higher-level agents oversee and coordinate lower-level agents—for example, in supply chain management, hierarchical agents monitor end-to-end global logistics, down to local deliveries. These agents coordinate shipping by strategically managing global routes, regional hub operations and individual package tracking across different organizational levels.
The Benefits of and Barriers to Adoption
This post is an update on ScheinerInc.’s September 9th, 2024 post, “Using AI Agents to Turn Thought into Action,” due to how rapidly AI continues to evolve and the growing interest in and deployment of AI agents. In particular, it focuses on how these agents are impacting the customer journey, streamlining processes while accelerating automation and integrating disparate workflows. A recent McKinsey post demonstrates how AI agents are impacting customer service and the financial benefits of using them: “Currently, the value of gen AI agents in the customer service environment is going to come either from a volume reduction or a reduction in average handling times. For example, in work we published earlier this year, we looked at 5,000 customer service agents using gen AI and found that issue resolution increased by 14% an hour, while time spent handling issues went down 9%.”
Granted, the risks and concerns associated with AI—pertaining to human oversight, privacy and ethical concerns, as well as data quality, employee distrust and cost of implementation—will continue to be barriers to driving adoption. However, I believe that the cost of not implementing AI, even in the smallest way, can negatively impact an organization’s productivity and its ability to innovate.
As always, thank you for reading or listening.
