Logo van gelder

Van Gelder Handles 44% More Customer Contact With the Same Team

hero-medium-full-customer-story-vangelder2

Scalable Service for 6,000 Customers a Day

yoursurprise-stats-customer-questions

More inquiries handled with the same team

by AI agents that take over repetitive first-line questions, so service staff can focus on contacts that need them.

yoursurprise-stats-wismo

Fewer delivery questions reaching the service team

thanks to a Delivery Agent that retrieves live route and order data and answers customers directly, without manual lookup.

yoursurprise-stats-chat-use

Reduction in phone volume as customers shift

to chat and WhatsApp, where AI agents provide instant answers to common questions.

appels

About Van Gelder

Van Gelder is a fresh produce supplier based in Ridderkerk, the Netherlands. They deliver vegetables and fruit to around 6,000 professional customers every day: restaurants, hotels, care institutions, and catering companies. With 5,000 products in stock and roughly 120 million euros in annual turnover, Van Gelder has grown from a family business into a highly mechanised operation where innovation has always played a central role, now driving their move into AI agents to scale service and improve the customer experience.

Story snapshot

Van Gelder's service team was handling a growing volume of contacts, with a large share of repetitive questions like delivery status updates that required manually looking up routes in a separate logistics system. Meanwhile, customers working in professional kitchens had little tolerance for waiting.

Van Gelder started with Mobile Service Cloud to centralize all contact channels, then added CM.com's personalization module for product recommendations. The next step was HALO. Starting with a Delivery Agent, Van Gelder now runs multiple AI agents in production. The result: 44% more customer contact handled with the same team.

The Challenge

Van Gelder's service team handled more than 75,000 customer contacts last year. A significant share were repetitive: delivery status questions that required manually looking up routes in a separate logistics system. Straightforward, but slow. Chefs in professional kitchens don't have time to wait on hold. Van Gelder also wanted to be more than a supplier. With kitchen staff shortages growing across their customer base, they saw an opportunity to offer product knowledge and recipe inspiration at scale.

If a competitor offers cucumbers for five cents less, we want our customers to say: Van Gelder contributes so much to my own success that I wouldn’t even consider switching partners.”

Robin Rietdijk, Marketing Manager at Van Gelder

gebouwvangelder

The Solution

Van Gelder's HALO setup started with a single agent and expanded from there. The platform connects to Van Gelder's logistics systems, product catalogue, and website, making real-time data available without manual lookups. Customers are reached via chat on the Van Gelder website, with WhatsApp as the next channel. Van Gelder deliberately started small: launching on chat first to monitor behavior, catch issues early, and iterate before scaling.

The Delivery Agent was ready in early December, but we chose not to go live before Christmas. We went live after the holidays, started monitoring daily, and when we updated the root agent to a new version, we caught unexpected behavior within hours because we were watching closely."


Robin Rietdijk, Marketing Manager at Van Gelder

Delivery Agent: Answering the Most Common Question Automatically

The most frequent question Van Gelder's service team received: where is my order? Answering it meant manually looking up routes in a separate logistics system. Since the Delivery Agent went live, 30% fewer of those questions reach the human service team. Next up: proactive notifications sent 15 minutes before arrival.

The agent handles this end-to-end, without human involvement:

  1. Identifies the customer and their active order

  2. Retrieves live delivery status and estimated arrival time

  3. Provides a direct answer and escalates when needed

Recipe Agent: Turning Product Knowledge Into a Service

Customers with limited kitchen staff sometimes need inspiration: what do I make with the tomatoes and leeks I have left? The Recipe Agent answers that at scale, generating tailored suggestions based on Van Gelder's full catalogue with direct links to the relevant products. Next step: connecting it to order history for proactive product suggestions.

The agent works as follows:

  1. Asks a few questions: cooking style, course type, etc.

  2. Generates three options and recommends one

  3. Builds out the full recipe and direct product links

  4. Allows follow-up for substitutions or adjustments

The Impact

Van Gelder handles significantly more customer contact than before, with the same team, faster response times, and human agents focused on conversations that actually need them.

yoursurprise-stats-customer-questions

More customer contact with the same headcount

yoursurprise-stats-wismo

Fewer delivery questions reaching the service team

yoursurprise-stats-phone-calls

Fewer phone calls as customers shift to chat

yoursurprise-stats-chat-use

Increase in WhatsApp conversations

Customers have shifted to chat and WhatsApp, sending a message between tasks rather than waiting on hold. Phone contact, previously the dominant channel, has dropped considerably. For a customer base where time in the kitchen is scarce, that change reflects a real improvement in the service experience, not just internal efficiency.

What’s Next?

The Recipe Agent will connect to the webshop to incorporate order history into its recommendations. The Delivery Agent will move from reactive to proactive with automatic pre-arrival notifications. WhatsApp, Van Gelder's highest-volume channel, will be activated once the chat agents are stable. An order modification agent is also in development, allowing customers to adjust orders directly through chat.

Lessons for Other Organizations

Van Gelder's entire HALO implementation was driven by two people alongside their regular responsibilities. A large team is not a requirement, but attention is.

Robin Rietdijk: "Start small and monitor closely. We launched on chat first, with lower volume, so we could see exactly what was happening. If you're not watching, you won't know."

Getting the underlying data and processes right before going live matters more than the technology itself. Once it's running, the work shifts to monitoring, improving, and building the next thing.

full-width-lessons-vangelder
Is this region a better fit for you?
Go
close icon