From agent to action: How to turn AI decisions into on-site personalisation

Cover image
Category: Use-case
15 May 2025

Predictive marketing is evolving. Platforms like Hightouch and Pega are making agentic AI more accessible, enabling brands to define in real time what each individual user should see, based on behaviour, product availability, and predicted intent. But even the smartest AI decision is wasted if it never reaches the customer it’s intended for. That’s where Contenzi comes in.

In this post, we’ll show how to combine agentic AI with Contenzi’s real-time rendering engine to turn intelligent predictions into live, on-brand website experiences using product data from Shopify.

How to integrate agents into the user experience

AI agents are increasingly able to decide what to show each customer. They look at signals like recent purchases, margin, stock levels, and behavioural trends to select the best offer or message for that moment. This kind of decision-making is rapidly becoming more advanced.

But making that decision visible to customers on the website is still a challenge. Most ecommerce teams are stuck waiting for developer support, or rely on static content blocks that do not respond to live data.

That is the gap Contenzi fills. While agents handle the decision logic, Contenzi is responsible for showing the right content on the site, in real time, using the latest product data and the brand’s own component library.

Example: Have an agent decide between a replenishment offer vs loyalty offer

A step-by-step example

Let’s walk through a realistic scenario based on replenishment behaviour. A customer regularly buys a certain type of dog food every 5 to 6 weeks. Your warehouse data has the full picture: past order dates, product sizes, margin, regional stock, and more.

1. An AI agent makes a decision

Based on historical purchase behaviour, stock levels, and margin, the agent selects the most relevant product and offer for the user. This decision is made available through a real-time input like the Hightouch Personalization API. Example response:

{
  "productId": "dogfood-123",
  “action_type": "replenishment",
  "message": "Time to restock your dog’s favourite meal?",
  "cta": "Reorder now and get 10% off"
}

2. Contenzi fetches the agent output

In Contenzi’s Data Reactor the decision is combined with content from Shopify like the product image, price and URL. This ensures everything displayed is accurate and up to date.

3. Contenzi renders the personalised experience

A visual block is displayed on the page, showing the offer exactly as selected by the agent. Contenzi takes care of layout, styling, and behaviour, using your own brand components.

A new way to personalise every visit

This approach does not rely on hardcoded variants or pre-defined audience rules. Everything is dynamic, driven by real-time decisions and real-time data.

You are no longer limited to static campaigns, pre-made content or one-size-fits-all journeys. Agents decide what to show. Contenzi makes it visible in real time, using live product info from Shopify. That means every visit can adapt to the customer, without adding complexity or waiting for developer support.

Ready to bring AI decisions to life?

Agentic AI is already reshaping how decisions get made in ecommerce. But those decisions only create value when they are delivered to the customer at the right moment, in the right format, seamless and on-brand.

Contenzi makes that possible. By connecting to the Hightouch Personalization API and integrating live product data from Shopify, Contenzi turns smart decisions into experiences that customers actually see.

This is not a future vision. It is something you can start testing today. If you’re exploring agent-based decisioning or already working with a solution like Hightouch or Pega, Contenzi gives you the missing link. It lets you show each customer exactly what your AI intended, without waiting for the next dev cycle.

Like this post?

We write about personalisation, composable architectures, and how to put your data to work in our newsletter.