AI
AI agent that handles 95% of customer questions, gives styling advice and increases the likelihood of purchase
Company: A fish named Fred
meet:
A fish named Fred
A fish named Fred is a menswear brand with bold prints and colour, recognisable in every item. In the webshop, it is all about choosing unique clothing. Visitors look at things like: is the size right, does it suit the occasion, and how does it combine with what someone already has in their wardrobe.
As a result, alongside practical questions about orders and returns, many styling questions also come in, for example about combinations, fit and when to wear something. A fish named Fred does not want service to become “support-speak”, but to keep the same tone and energy as the rest of the shop. That calls for answers that are clear, stay on brand and at the same time help commercially with the next step towards a purchase.
the challenge
A fish named Fred deliberately wanted to be ahead of the market and be one of the first to test how an AI agent adds value in the webshop. Many recurring questions also came in about orders, returns and product choice, questions that lend themselves well to smart support.
At the same time, we saw an opportunity to connect service with commerce, without it feeling forced. A logical side benefit was that the AI agent could also give styling advice, in keeping with the distinctive brand style, so conversations could more often move towards a concrete choice and purchase.
The brief was to build an AI agent that supports customers in the brand’s tone, handles standard service questions directly, and guides conversations naturally towards a suitable product choice and purchase.
our solution
We started by defining the brand and tone of voice, with fixed guidelines for language, form of address and limits on advice. After that, together with the team, we set up a knowledge base with product information, return conditions, the ordering process and frequently asked questions, so the AI agent can answer consistently and on brand.
On top of that, we added targeted logic that connects service and commerce:
Relevant product recommendations based on the question and context.- Styling tips when customers are unsure about combination, occasion or fit.
- Clear routing to the team for exceptions, so the agent does not force unnatural advice.
- Making online marketing scalable without costs getting out of hand
Next, we implemented the AI agent on the website, including a fallback to the team for exceptions and more complex questions. After going live, we optimised the conversations based on chat data, by adding knowledge and adjusting answers where customers asked follow-up questions or got stuck.
the result
The combination of one recognisable employer story, target group profiles and a measurable funnel approach gave more grip on recruitment marketing. Kiwa saw not only growth in visibility, but also more clearly which target groups responded and via which route most applications came in.
94%
of conversations are handled entirely independently by the AI agent.
But also:
- less pressure on customer service, because most standard questions are handled by the AI agent.
- styling advice is part of the conversation, so the service fits the brand better and more often leads to suitable products.
- upsell through product recommendations has visibly increased, based on the recommendations made and followed up in conversations.
The next step is an integration with Shopify, so the AI agent can give order status, shipping information and return updates directly in the chat.
what the client says
“Valuable partner”
“BE Digital has helped us take our digital approach to a higher level. As a specialist, they think along at a strategic level, but they also make sure that concrete improvements are made to our processes, partly through AI. A big plus is the knowledge transfer to our team. Through targeted training for marketing and sales, BE Digital has made sure our colleagues better understand how digital contributes to growth, efficiency and customer-focused working. In a time when digital and AI are becoming ever more important, it is crucial to invest in this structurally.”
Rob Schalker
Owner A fish named Fred
Owner A fish named Fred
the BE Digital Touch
From the first moment, we did not approach this as “putting a chatbot live”, but as a controlled test to learn faster than the market, without risk to brand or service quality.
We began with tight guardrails, tone-of-voice rules, scope definition and fallback, so the agent only answers what is certain and passes everything outside that on to the team. After that, we made it measurable with one clear definition of success and steered based on real chat data, not on assumptions or scattered feedback.
We began with tight guardrails, tone-of-voice rules, scope definition and fallback, so the agent only answers what is certain and passes everything outside that on to the team. After that, we made it measurable with one clear definition of success and steered based on real chat data, not on assumptions or scattered feedback.
The extra something lay in the combination: the same agent handles service questions, but also recognises moments where styling advice and product recommendations make sense, so a conversation does not end at “resolved”, but more often ends in a concrete choice and purchase. By setting up knowledge management as a fixed process, with recurring updates and ownership, quality became predictable and the agent stayed on brand, even as questions and range change.
knowledge management as a fixed process
We work with, among others:
