This is the second blog in Yali Sassoon's Agentic AI series. Click the links below to access the other blogs: ‍

• Agentic Applications: Transforming Customer Journeys for Brands

‍

Over the last year we have seen a paradigm shift in artificial intelligence. AI-powered agents such as OpenAI’s Operator have emerged. These agents can browse the web and perform tasks on behalf of users—your customers. If your brand’s website isn’t optimized for these new digital assistants, you may soon be left behind.

In this post, we’ll look at:

• What AI-powered agents like OpenAI’s Operator actually do
• How AI Agents have changed the engagement game for brands
• How brands can start optimizing their websites and apps for an agentic world
• The concept of Agent Experience Optimization (AXO) and why it matters

How AI Agents Will Transform Web Engagement

Agents will become the new web browsers. Instead of personally clicking through dozens of sites, consumers will simply instruct an agent to “Find me the best Italian restaurant near Oxford Circus tonight,” or “Buy my weekly groceries.” The agent will visit various sites, compare options, and carry out the final transaction—often without requiring further human input.

OpenAI’s Operator: A Glimpse of the Future

Earlier this year, OpenAI announced Operator - an agent that can navigate the web and perform tasks on behalf of an end user. As a consumer, you or I can ask Operator to perform a range of tasks on your behalf.Operator will then go and do them for you, reporting back later. To give just a few examples of things Operator can do:

1. Make a dinner reservation 
2. Book a taxi 
3. Buy your groceries 

In each case, the consumer provides Operator with an objective - e.g. “Please book me a dinner for four in Soho that starts sometime between 7.30 and 8pm. Ideally an Italian restaurant that is highly rated on Yelp, and within walking distance from Oxford Circus.” 

Operator then works to fulfill this request, but will come back to the consumer if it needs help. (Maybe it’s not sure about which of two restaurants to pick, or needs the consumer to provide her credit card details.) After a short period, Operator will complete the tasks and report back.

While Operator is only in a research preview and has not yet been widely rolled out, it provides us with a clear view of how consumer interactions with websites and mobile apps are likely to change. As I argued in my last post, there is a raft of consumer workflows that are ripe for reinvention with agentic applications. 

What I did not say there, but will say here, is that the field is open for anyone to go and build those agents: you do not necessarily need to be a travel brand, for example, to build an agent that books a holiday on someone’s behalf. (Though I think it will help.) And we should certainly expect a raft of major tech companies, from OpenAI and Anthropic, to Microsoft and Facebook, to start offering these kinds of agents.

This Changes… Everything

Brands have spent decades optimizing their websites and applications for humans, leveraging techniques from Digital Experience (DX) through programmatic display advertising to Search Engine Optimization (SEO).

But now, if you’re a retailer - soon in addition to real customers, you will have a growing number of AI agents on your site buying on behalf of your customers. For media sites, it won’t just be customers reading content; you will have agents collating content on behalf of your readers (and then compiling and condensing it on behalf of those readers). If you’re a financial services company, you won't just have customers researching financial products or processing payments, but you will likely have agents doing so on behalf of your customers.

This has profound implications for brands.  After almost three decades of optimizing online experiences for humans, they now need to think about optimizing those same websites and applications for agents.

Why Brands Need to Optimize for AI Agents

Today, brands spend a lot of effort on optimization to drive their commercial goals. In reality, this means optimizing for customer (human) experience. For example:

• Customer Acquisition: Managing spend across different marketing channels and campaigns to drive people to their website
• Conversion Rate Optimization: Managing the buyer journey so it is as streamlined as possible, driving people to complete their purchases
• Promotion and Pricing Optimization: Using customer intelligence (built on customer data) to optimize promotions and pricing on different products to maximize profit

These activities need to be reimagined for a world where agents are major consumers of websites and applications. Let’s take an ecommerce store example:

• Customer Acquisition: How do we make sure that when a customer tasks an agent with finding a product we sell, the agent knows to check for that product on our website? How do we encourage and incentivize the agent to use our website rather than our competitor’s?
• Conversion Rate Optimization: How do we ensure that the agent can find what it is looking for on our website, and once it does, proceed quickly and easily to checkout and transaction? Note if an agent repeatedly struggles to do this, and finds it easier on our competitor site, it is likely to stop bothering using our website in the future at all
• Promotion and Pricing Optimization: How do we optimize our merchandise for agents? Is there any way we can understand the instructions an agent is acting on, and adjust pricing and promotion accordingly? Or are different strategies more appropriate for agents? (Fixed price, or sharing our algorithm with the agent?)

Introducing Agent Experience Optimization (AXO)

Agent Experience Optimization (AXO) means designing digital properties so that AI agents can seamlessly discover, assess, and transact. It involves:

• Identifying AI Agents: Knowing how to distinguish agent-driven traffic from human customers or malicious bots
• Analyzing Agent Journeys: Tracking whether agents effectively find products, add to cart, and check out — and pinpointing where they get stuck
• Facilitating Agent-Human Collaboration: Understanding how agents and humans cooperate together to reach goals on behalf of the user, and opportunities to provide a better canvas to enable an effective customer-agent collaboration

Brands that want to keep serving these customers need to do Agent Experience Optimization. They need to understand (to the extent possible) how agents make decisions about which websites and applications to visit to meet different types of customer needs, and take any possible actions to make it more likely that they will be included. 

AXO is Not Just the Next Generation of SEO

You may be thinking that there are some interesting parallels between Search Engine Optimization (SEO), where brands optimize their digital experience for organic search, and Agent Experience Optimization (AXO), where brands optimize their digital experience for agent engagement.

The parallel is intriguing - but don’t fall into the trap that AXO is just the next generation of SEO. Remember that brands are furiously building their own agents to augment and replace the current digital experience for humans. Because brands have granular consumer history and cross-customer behavioral benchmarks, they can super-charge their agents to be more personalized to their specific brand than Operator ever can.

Put AXO together with the rise of brand-owned agents, and much of the focus of AXO will be on mediation – even negotiation – between consumer-owned agents and brand-owned agents. Expect more on this in a future post!

How to Get Started with Agent Experience Optimization

At Snowplow, we’re building infrastructure to help brands track and understand agent interactions on websites and applications. If you’re ready to explore Agent Experience Optimization — and stay ahead of the AI revolution — get in touch to learn more.