Finding the ‘next Max Verstappen’ with data and smart algorithms.
Van Amersfoort Racing
Recommend.ai
Recommender voor datagedreven aanbevelingen.
Use customer, product and transaction data to generate relevant recommendations
Trusted by leading organisations
Why we built this,
Customers often see the same products, content or offers, even though their interests and past behaviour differ. Prices are also frequently set using fixed rules or based on experience.
As a result, what is offered does not always match what an individual customer is looking for or the current commercial situation. With large product ranges and many transactions, it is also impossible to set every recommendation or price manually.
The Recommender uses available customer, product and transaction data to generate relevant recommendations. These recommendations can become increasingly relevant and personalised. This helps our clients improve the user experience, increase engagement and help their customers make the right choice more quickly. The application can be configured for different settings, from online shops and retail to media and mobility.
What you can expect,
Recommendations tailored to the context
Recommender analyses available data to determine which product, item of clothing, video, image or message might be relevant to a customer or user.
The recommendations you can offer depend on the available data, the range of products and when the recommendation is shown.
Hybrid recommender system
The solution combines content-based filtering, which makes recommendations based on characteristics of previous preferences, with collaborative filtering, which uses patterns from similar users to predict new interests. Combining both techniques in a hybrid recommender system makes the recommendations even better
Decide which commercial rules apply
A model does not have to decide entirely on its own what a customer sees or pays. You can specify which products may be recommended, what price limits apply and when an employee needs to review an outcome.
For example, you can configure it based on:
- Availability of products or services
- Defined price limits
- Commercial priorities
- Relevance to the user
- Exclusions and other business rules
The available rules and configuration options for Recommender still need to be determined.
Our approach,
-
Start with one specific decision
We first identify the decision point to focus on. This could be a product recommendation, the next piece of content or a pricing recommendation.
We then map out what data is available and how the organisation currently decides what to offer. -
Develop and test a model
We develop a model suited to the problem and test its outputs against historical or representative data. We compare its recommendations with the existing approach.
We assess not only how the model performs technically, but also whether its outputs are useful within the commercial process. -
Test and expand in a controlled way
We start with a defined product range, customer segment or channel. We monitor the results before expanding the application further.
New products, channels or decision points can be added to the same solution later.
Not arbitrary personalisation, but a transparent tool that supports commercial decisions with data.
This is how Recommender supports commercial decisions with data, while your organisation determines which rules and limits apply.
What we've built before,
Three projects from our work with data and AI.
Further reading
Blogs and in-depth articles from our team.
24 Aug 2026
Three lessons from our AI projects
What we learned from dozens of AI projects: start with the decision, treat data quality as an organisational issue and arrange management in advance.
18 Aug 2026
From pilot to production: why most AI projects stall
Most AI projects stall not because of the technology, but because of data, teams and processes. Read what it takes to move from pilot to production.
25 Jun 2026
Responsible AI: from principles to practice
AI is developing rapidly. New models emerge almost every week, and more and more organisations are experimenting with AI.
Get in touch
Stef Stoelwinder
Business DevelopmentE-mail info@datasciencelab.nl
Phone +31 (0)20 3636 724
Frequently asked questions,
The application can be configured to recommend different things, such as products, clothing, content, images or messages. Which recommendations are relevant depends on what you offer, the point in the customer journey and the data available.
That depends on the application. Transactions, interactions, product information and available context can all play a role. Not every solution needs to use all available customer data. We determine which information is necessary and which data can responsibly be made available.
The solution can support pricing recommendations when sufficient relevant data and clear commercial conditions are available. Whether prices are adjusted automatically depends on how you want to configure it and the risks involved. The exact functionality for dynamic pricing still needs to be confirmed.
A model based solely on past behaviour can reinforce existing preferences too strongly. That is why its configuration should also account for variety, new products and other commercial goals. The right balance depends on the application and the user.
It starts with a clear goal. Depending on the application, you can look at usage, conversion, margin or how often recommendations are followed. A controlled test shows whether the model performs better than the existing approach.