AI agents are self-governing software applications. They use artificial intelligence to solve problems, make decisions, and perform tasks on behalf of customers, healthcare patients, and other users. Also called agentic AI, AI agents control their own multi-step processes. They can adapt to circumstances, context, and data in real time, then take action with minimal human intervention.
AI agents are on the rise, and agentic AI is transforming how brands can engage with customers at a time when the customer journey is fragmenting. For instance, an AI agent has the ability to predict individual customer behavior and aggregate it into larger trends. Brands can then use AI agents for everything from automated campaign management to next-level personalization.
How AI agents are different from AI search engines
AI search engines and AI-driven search platforms are built to generate information and share it conversationally as they answer customer questions.
AI agents can retrieve and generate answers to customer questions. Then, they can act on the data they surface and share it with customers.
An AI agent, like an AI search engine, can help patients "find a dermatologist nearby who takes Optum insurance and specializes in microdermabrasion and adult acne treatment."
The AI agent can go to the next level, though. After it helps customers find healthcare providers (or restaurants, shops, financial advisors, etc.), the AI agent can automatically book an appointment for that customer. In the case of the customer looking for a dermatologist, the AI agent might also collaborate with them to design an inflammation-reducing meal plan and a skin care regimen. Then, if it's been set up with the right parameters, it might purchase skin care products, automate seasonal facials, and arrange for grocery delivery based on the meal plan.
The basics of how agentic AI works
While a customer "queries" or "prompts" an AI search platform, customers "partner" or "collaborate" with AI agents. And brands definitely want to surface in those collaborations.
There are three main types of collaboration:
1. Defining goals and parameters: The customer gives the AI agent instructions or a high-level set of objectives. Then, the AI agent interprets them and uses them to inform its behavior.
Financial Services Example:
"I want to update my financial portfolio so it's more conservative. When interest rates hit X%, there are significant disruptions in the global supply chain, pandemic level indicators emerge, or a nation-state in the West starts war mongering, send me an alert and advice on how to respond. Include recommendations from Vanguard, Schwab, and Scott Galloway. Show me links to all the sources you're basing your advice on, too, please."
2. Delegating tasks and creating workflows: The customer gives the AI agent assistant an assignment, often defined by multi-step processes and complex considerations to achieve a goal.
Hospitality Example:
"Consult my personal and work calendars over the next six months and look back at them for the last two years. Then, help me find at least two 10-day windows this year that may be ideal for a vacation. (No product launches. No recurring doctor appts, etc.) With those windows in mind, plan three potential trips: one to the UK, one to SE Asia, and one to the American West. Focus on intermediate-level hiking, beaches or rivers, and natural areas. I want to stay in Airbnbs with 3.5 star reviews or higher. Include potential flights departing and returning to Toronto (Munro or Pearson are both fine). Red eyes preferred, at least for the first and last leg of the trip."
3. Collaborating in real-time or at key intervals: The customer and the AI agent work in tandem, and the AI agent assistant adapts its actions based on feedback from the customer and/or the changing conditions surrounding their collaboration.
Retail and Direct-to-consumer Example:
"Here's a mood board. I want to create a vendor list and purchasing plan for furnishing my new 800 sq. foot apartment. I'm fine with a hi-lo mix of furniture and accessories, but I want to have the option to see each item in person before I buy. My budget is flexible, but let's target $12,500. Organize it in a table format, please. Not a list. Keep the thumbnails, product links, and prices visible. Now turn it into a pivot table with columns for each room: Dining, Kitchen, Bedroom, Living/Workspace. Style is on point. But less veneer and velvet. More easy care, pet-friendly in-stock fabric options. Can we mix in a few more budget-friendly items, especially for bedroom furniture and desks/case goods?"
AI agent examples and AI agent assistant use cases for brands and customers
AI agents can work like personal assistants, called AI agent assistants. They can help customers discover brands, research products, plan trips, book appointments, and offer personalized advice. AI agents can analyze past purchases, browsing behavior, and location/travel data. Then, they can recommend products, services, and brands across digital touchpoints. They can even influence behavior when the customer travels near a brand's brick-and-mortar location.
Brands can use AI agents to analyze customer sentiment and predict behaviors. They can also influence customer sentiment and behavior with more personalized campaigns with relevant, enhanced content. They can even activate dynamic pricing and inventory updates to test customer intent and drive conversions.
How AI agents use knowledge graphs to operate intelligently
The relationship between AI agents and knowledge graphs is symbiotic. Knowledge graphs provide data frameworks for AI agents. Composed of structured and unstructured data sets, knowledge graphs are like a brain filled with information. AI agents will mine, understand, and contextualize knowledge graph data when they work with customers.
AI agents are like the synapses in the nervous system. They connect the brain (or knowledge graph) to the signals that a customer using the AI agent sends. AI agents take the data in the knowledge graph and adapt to it, so it can meet the customers' goals and requests for delegation or collaboration.
AI agents interpret the knowledge graph with machine learning, NLP, LLMs, and other intelligence models. Then, they activate it. Sometimes, those actions are recommendations and insights. Sometimes, the customer has empowered the AI agent to make decisions or take action on their behalf.
Like a brain, knowledge graphs act as a brand's single source of truth. Yet, they're easy to update with new information as brand and customer data sets grow. Naturally, the better your knowledge graph, the more you can impact the quality of interactions between your brand, your customers, and your AI agent.
Pro tip: With Yext Knowledge Graph, brands can give AI agents a comprehensive, informed, and stable data foundation to work with. Yext also gives brands the power to automate complex, manual workflows (like hundreds of listings updates, reviews management, or local social media). This improves team efficiency and creates better, personalized customer experiences.