Agentic AI Applications: How They Will Turn the Web Upside Down
Websites today are designed around the goals of the enterprises that own them. The next generation of agentic websites will reflect the needs (and contexts) of the users visiting them.
A month ago I published a post looking at how agentic applications were going to transform customer journeys. In this post I’m going to relook at why and how agentic AI applications will transform the web more broadly, by enabling brands to build websites that better understand the unique intentions of each of their website visitors, and respond to deliver them a much more appropriate experience.
The web is going to change dramatically, for the better, for the vast majority of people on it.
The web today is fundamentally broken
Don’t believe me? Consider an ecommerce site. There are many reasons someone might visit an online retailer’s store.

However, the website that the visitor lands on has, typically, been designed primarily to drive her to buy products. If the visitor wants to do anything other than tasks involved in the buying process (e.g. research the retailer, apply for a job, get product support etc.) she will need to figure out how to navigate to some small part of the website that caters to her need to accomplish her goal.
The design of the website reflects the goals of the retailer: namely, to sell stuff, rather than the goals of the visitor. For the subset of users who have an intention that aligns with that goal, the digital experience is reasonable. (It’s not as good as it could be because it’s not as specific as it could be: most retail stores are geared around selling all their products, or at least all their highest margin / in stock / best selling products, whereas most customers who visit a store have the intention to buy only one or a handful of products.)
Everyone else has a rubbish experience - they have to work hard to get off the “mainline” funnel that the website is optimized to push them through to find the information and functionality they need to complete their specific task.
The result is 98% of a retailer’s store is geared towards selling stuff, but only 2% of visitors buy! Something is clearly wrong.
This problem is not limited to ecommerce. You see it in every B2B SaaS website.
These websites have been designed to take an individual, who has a limited or no knowledge set about the SaaS product, and provide enough information and compelling calls to action that they will sign up for a trial or speak to an SDR to learn more.
They are not geared towards anyone else, for example, existing customers who want to dive into specific new use cases, or former customers who want to find out how the product has evolved in the last couple of years. These visitors have to work hard, navigating around the site and translating the content to answer their set of questions.
If anything, the problem in B2B is even more acute. Business buyers of SaaS solutions each come with their own set of problems to solve, their own (different) understanding of the problem domain and their own set of constraints.
Even if two visitors are both there to learn and potentially buy the product, they might have wildly different questions to answer and steps they need to take. Each of them will have to wrangle the same static UI to get the information they require.
Agentic AI applications offer a customer-centric future
Agentic AI applications for the web offer the possibility of an alternative future, one that is oriented around the visitor not the brand
If an AI agent understands a visitor’s intention, it can create a bespoke experience just for that visitor. So if the visitor asks the agent “how do I return this product?” - the agent can provide all the necessary information.
Better still, the agent should be able to look up the customer, the particular order, and their location, and provide very precise instructions, making the return as simple as possible for the customer to process.
If on the other hand the customer is interested in browsing a particular product category, the agent can curate a magazine-like experience optimized for the customer’s device, and personalized according to the customer’s tastes.
Agentic AI applications have a range of capabilities that mean that they are able to create unique, customer-specific experiences oriented around each individual customer’s intentions and preferences. They are able to:
- Triangulate customer intent from multiple different customer signals, including what the customer has explicitly told the agent and how the customer is behaving right now. If the agent is not sure, it can ask the customer outright what she wants to accomplish, or explain what it thinks the customer wants, and ask the customer to confirm
- Use tools to fetch customer intelligence, so that they can quickly build an understanding of who this customer is, their brand preferences, price sensitivity, shopping style and other key facts that can help deliver a compelling, customer-centric experience
- Use large language models (LLMs) to dynamically generate specific experiences for specific customers. (Because LLMs can generate text, images, video and code.)
- Reflect on how the customer is responding at each stage in the customer journey, and update the next step in the journey accordingly
Will brands adopt customer-centric approaches?
While this capability exists, will brands use it to deliver customer-centric experiences?After all, brands could still choose to build agentic AI applications oriented around their goals rather than the goals of their customer.
The reason I’m optimistic that this will not be the case is that:
- As I argued before, anyone can create customer-facing agentic experiences. Customers are going to engage with AI agents that orient around their goals, not their suppliers, so the competitive pressure to build systems that are customer-oriented will be immense
- Brands know that the most profitable customers are the happiest customers. The reason that today’s web is not customer-oriented is that customer needs are very varied, and without agentic AI technology it’s very hard to be able to identify and orient around those different needs.
The future for consumers and brands
As consumers, we should all be excited about this future. AI-powered agents will help us accomplish specific tasks more efficiently.
The next generation of websites will be agentic AI applications. We’ll be able to tell these applications what we want, and they will use that information to give us exactly what we want. No longer will we have to wrangle a website designed to meet one set of needs to try and get it to meet our needs. The AI agent will do that work for us.
Brands should be excited about this future too
Brands will be able to engage customers much more effectively when they are on their websites i.e. when they want to engage with the brand.
They will be able to use AI agents to have meaningful conversations with their customers at scale, building a much better understanding of those customers and delivering each of them a markedly better experience.
Retailers will no longer look to drive conversion rates from 2.1 to 2.3%. Instead, they will be looking to understand why those people who came on their site to buy didn’t buy, and making sure that we use that intelligence to make sure that when customers with similar needs visit in future they’re provided with a much better service, and are that much more likely to convert.
So no more 2% conversion rates: those rates will be 20% or more.
Want to get started building agentic AI applications for your customers?
At Snowplow we’re really excited about the next wave of agentic customer-facing applications. We’re building the Customer Data Infrastructure to enable brands to build those experiences. (I’m going to be blogging more about exactly what we’re building and why shortly.) If you’re interested in learning more then get in touch!
Credits
Two posts were really helpful for me in formulating my thoughts here
- Scott Brinker’s excellent post on Systems of Context and Systems of Truth
- Venk Chandran at Pathfactory pointed me at this excellent report on the impact of agentic applications in the B2B buyer journey