Case Study
A custom AI agent built for an automotive company in the Netherlands, set up to handle the repetitive enquiries and routine follow-ups so the team can stay on the tools instead of the inbox.
By Rens Schulte, Founder & Lead Engineer
The Problem
A custom AI agent for the work that kept stealing the day.
When Schulte Automotive came to me, the problem was simple to state and easy to feel: the same questions and follow-ups kept landing all day, every day. Each one was quick on its own, but together they pulled the team off the tools and into the inbox. I set one honest goal with them. Build a custom AI agent that handles the repetitive enquiries and routine follow-ups in the way they already work, so the team can stay focused on the cars and only the real exceptions reach a person. No off-the-shelf chatbot, no rigid script to fight later.
The Approach
I built the agent from scratch on Claude, shaped around how Schulte Automotive actually answers customers. No off-the-shelf chatbot, no rigid decision tree, and no generic script that breaks the moment a real question arrives.
I fit the agent into their existing way of working rather than asking them to change it. It picks up the routine enquiries and follow-ups where they already happen, so adopting it felt like less work, not another tool to learn.
The agent handles what it can confidently handle and hands anything unusual straight to the team, with context attached. No customer gets stuck in a loop, and nothing important slips past unseen.
I set the agent up to answer only what it can stand behind and to flag the rest, so it stays accurate and on-brand. A dependable foundation I can keep tuning as their needs grow.
How I Worked
I started with a short, focused conversation about the enquiries that fill their day and how they answer them today. From that I agreed a clear scope for the agent before any build work began.
I built the custom Claude-native agent around their real questions and tone, tuned so its answers sound like the team and not a generic bot. Reviewed on real examples before going further.
I wired the agent into how they already work, set the handoff rules for anything unusual, and tested it against real enquiries so it behaved the way they expected everywhere it ran.
I switched the agent on, ran a final check and handed it over, with the client owning the result end to end and a clear path to keep tuning it.
The Result
The result is an AI agent the client is genuinely happy with: it quietly handles the repetitive enquiries and routine follow-ups, so the team spends less time in the inbox and more time on the work that pays. Communication stayed straightforward from the first call through to launch, and the agent went live on time.
These describe how the agent works, not invented outcomes. I report what it reliably does and stand behind, never made-up time-saved or revenue numbers. A real measured figure goes here once the client confirms it.
“Rens understood what we needed and got it right. Quick to respond, clear about every step, and we are very pleased with the result.”
Let's connect
Want an AI agent like this for your business? Send a short brief and you'll have a written quote within 24 hours. No drawn-out discovery sessions, no surprises later.
Where we work
A solo studio working with growing businesses across the Netherlands. Projects run fully remote, built and run end-to-end by the founder, Rens Schulte.