Start a project
All cases

Getting more out of the customer-service tools already in place

Instead of adding another support platform, we focused on making De Bonte Koe’s existing Trengo setup do more, starting with repetitive work the team was still handling manually.

  • Customer service
  • AI automation
  • Trengo
  • Shopify
Client
Duration
2 months
Led by
Miloš
Category
Automation
An opened De Bonte Koe gift box: a tote bag, a coffee cup, chocolate and branded packaging.~2 hof manual labelling saved every week

At a glance

What we did

  • AI ticket labelling
  • Support categorisation
  • Chatbot testing
  • Shopify support integration
  • Voice consolidation planning

The challenge

De Bonte Koe already had a capable customer-service platform, but part of the team’s time was still going into repetitive work that the system could handle automatically.

The first priority was straightforward: reduce manual ticket labelling without taking control away from the support team.

How we helped

  1. 01

    Built the automation from real support history

    We analysed 3,310 previous tickets and used the recurring patterns to create a two-level classification system covering five main support categories and ten complaint subtypes.

  2. 02

    Automated the repetitive work first

    The AI labelling system now handles a task the team previously did manually. Trengo’s own time-saved analytics showed five hours saved over the latest two-week period, closely matching the team’s real-world estimate of around two hours per week.

  3. 03

    Built the next improvements around the same system

    A customer chatbot is currently being tested with clear human-handover rules, while the team is also consolidating more of its support stack into Trengo instead of paying for overlapping tools elsewhere.

The Cowbot chat on the De Bonte Koe site: a customer asks where to buy chocolate mousse.
Cowbot answers with the shops and weekend hours where the mousse is sold.
Rows of envelopes sealed with orange De Bonte Koe stickers.

The outcome

Manual ticket categorisation has largely disappeared, saving the support team around two hours every week while creating cleaner, more consistent support data.

The broader result is a customer-service setup that gets more value from software the business already pays for instead of solving every new problem with another app.

Insights

  • Get more from what you already pay for.

    The existing Trengo setup could do more before anything new was added.

  • Let real history shape the automation.

    3,310 past tickets defined the categories, so the labels follow real support patterns.

  • Start with the repetitive work.

    Labelling came first; the chatbot followed, with clear rules for handing over to a person.

“Would highly recommend Milos. Fast learner, can operate in non structured/growing environment.”

Amy KleinChocolate Executive, De Bonte Koe