Cases — Agentic AI projects
Agentic AI cases with ANWB, AFAS, InShared, Wolters Kluwer and more. Concrete AI agents and automation delivering measurable value in production.
Proven results
An honest look at the work we do for clients — what we built, why, and what it actually changed for them.
All
Insurance
Legal Tech
Accountancy
Retail
Knowledge assistant for member services
ANWB
Mobility
Agents type a question and get the right passage from the knowledge base back, with the source attached. Saves a lot of clicking around and second-guessing.
- Faster call handling
- Consistent answers
- Less manual searching
Asking your data questions in plain English
Wolters Kluwer
Legal Tech
Business users just type their question in normal language. Behind the scenes a semantic layer makes sure the answer comes from the right, governed dataset.
- Faster data access
- Explainable data queries
- Fewer ad-hoc requests
A writing helper that lives inside the case file
InShared
Insurance
No separate window, no extra tool. An assistant inside the case system drafts emails using real customer data and sticks to the house style. The agent reads it, tweaks it, sends it.
- Faster email drafting
- Consistent tone
- No context switching
Mail that sorts itself onto the right pile
Meijers Verzekeringen
Insurance
Around 150 emails come in every day and all need to go somewhere. A workflow in their own cloud environment now sorts and routes them. We tested it quietly first, then flipped it on.
- Half FTE freed up
- Faster mail routing
- Secure local processing
One place for learning and growth
AMI Kappers
Retail
A single platform where e-learning, onboarding and personal development conversations come together. Connected to HR data, with dashboards per role and a learning assistant that thinks along.
- Central knowledge platform
- Standardised onboarding
- Clear HR overview
A set of agents that handle marketing work together
AFAS
Software
An internal platform where multiple agents collaborate on research and content for the marketing team. We set it up as an open framework, so they can keep building on it themselves.
- Scalable agent framework
- Automated marketing research
- Extensible code base
Reading supplier rate sheets without the headache
TravelEssence
Travel
Messy supplier rate sheets go in, a workable matrix comes out. AI does the reading, fixed code does the maths, a human signs off on the result.
- Minutes instead of hours
- Reproducible calculations
- Mandatory human review
Applications that prep themselves
Domus Valuas
Healthcare
Incoming application emails get read, attachments handled, and a forwarding email is waiting as a draft. Sending stays with HR — we kept that on purpose.
- Faster application processing
- Automated admin routine
- HR retains final control
Our own research agent
OpenClaw
Internal
The field moves faster than any one person can follow. An internal agent scans new frameworks and models continuously, filters out the noise, and only flags what actually matters to the team.
- Continuous tech monitoring
- Faster tool adoption
- Structured team reporting
From scattered annoyances to a working stack
4Everyware
Trade
We took on the manual work around leads, data and invoices with modular pipelines and hooked everything into the existing CRM. People now work from one portal instead of six tabs.
- Reduced daily manual work
- Company-wide AI platform
- Modular infrastructure setup
An honest advisory report across four departments
Beachclub Indigo
Hospitality
We sat in with admin, sales and marketing, wrote down what actually got in the way, and turned that into a roadmap at three maturity levels. Not an 80-slide deck — something they could start on Monday.
- Concrete automation roadmap
- Departmental quick wins
- Phased implementation plan
BKSM: ditching the shared inbox for a real ticket system
Monarch
Industrial
Their shared Outlook inbox was drowning. We replaced it with an on-prem ticketing system that scores urgency itself and assigns mails to the right person. Data never leaves the building.
- Eliminated manual triage
- Live SLA overview
- Local data processing
Invoices and orders straight into Microsoft Fabric
The Onion Group
Agriculture
Supplier invoices and call-off orders get read, validated, and pushed via a secure connection straight into the ERP and the data lake. Nobody has to type them in anymore.
- Eliminated manual data entry
- Centralized Microsoft stack
- Automated ERP imports
A web shop chatbot you can trust at 2am
Kitchen 247
E-commerce
A chatbot on the web shop answers product questions day and night with current product info. Underneath sits a deployment pipeline that tests every change before it ever goes live.
- Continuous customer support
- Reduced support load
- Fast and safe deployments
FAQ chatbot that nudges visitors toward the showroom
Kuchenwelt
Retail
A chatbot that answers 125 specific FAQ questions on the spot and gently guides browsing visitors toward the showroom. Improves as the knowledge base grows, runs on the same CI/CD stack.
- Faster answers for visitors
- Less load on support
- More showroom visits
Absenteeism reports in minutes instead of a day
Van Campen Consulting
Consulting
Excel goes in, a finished Word report comes out. Two cooperating models keep their methodology and tone of voice in check, so every report reads like it came from the same advisor.
- Faster report generation
- Consistent house style
- Higher analytical quality
A tidier inbox with templates they manage themselves
Hizi Hair and Kinki Kappers
Retail
Incoming mails are read and labelled, and for standard questions a draft reply is ready to go. The templates live in an Excel the client manages directly — no developer needed to tweak a line.
- Organized shared inbox
- Less manual typing
- Client-managed templates
Smart Planner and tenant tickets during the Yardi migration
Green Real Estate
Real Estate
A daily meeting-room schedule on SharePoint, plus a flow that turns tenant emails into proper work orders for Yardi. Fixing two things at once during a migration is no small feat.
- Daily time savings
- Smooth ERP transition
- Scalable n8n foundation
Custom Claude skills, one employee at a time
Van Deursen Group
Real Estate & Investments
Over six weeks we built AI skills tailored to each person's actual job — from digging through quarterly fund reports to calculating WWS points. Not a generic training, but something that fits their role.
- Less repetitive work
- Time saved per employee
- High adoption rate
Meeting intelligence that stays inside the EU
Based
Consulting
Meeting audio gets transcribed and turned into structured reports. The whole thing runs on EU infrastructure and meets GDPR — a requirement that's rarely negotiable in consulting.
- Searchable meeting database
- Full GDPR compliance
- Better knowledge retention
M&A purchase agreements filled in from Chamber of Commerce data
Act Legal
Legal
We put existing software next to a custom plugin and had both auto-fill contracts from Chamber of Commerce data. The result gave the team a clear basis to make a call.
- 80% time savings target
- Legora OS 94% accuracy
- Plugin 89% accuracy