← All workIntelligence

Wtp tooling for pension funds.

ClientA Dutch pension fund
RoleDesign, build & run
StackNext.js · Supabase · multi-model AI
Statusv7.5 in production
Wtp tooling for pension funds
01The challenge

Nine targets, 1.3 million people, one new pension law.

The Dutch pension transition (Wet toekomst pensioenen) asks funds to prove their participants actually understand the new system, not vaguely but against hard knowledge targets with deadlines. For the communication team that means tracking nine objectives across 1.3 million participants, with the evidence scattered over survey tooling, web analytics, search data and a printed magazine. Nobody has time to stitch that together by hand every quarter.

02The build

We built the team a single place where the answer already lives: a dashboard that tracks every objective against its target, with an AI analyst on top that speaks Dutch, knows the data and renders its own charts mid-conversation. Fresh numbers flow in on their own: web analytics daily, search volumes weekly, magazine scans monthly.

Objectives dashboard with live target tracking
AI chat analyst with generative UI: charts render inside the answer
Automated syncs: analytics daily, search data weekly, magazine QR monthly
Admin panel for indicators, events and chat analytics
Multi-model AI layer (Claude, GPT, Mistral), swappable per question
Self-hosted database on EU soil, GDPR-first by design
03The outcome

From quarterly guesswork to a daily answer.

The communication team now opens a dashboard instead of running a spreadsheet ritual. Progress on all nine objectives is visible the moment the data lands, and every question in between gets answered in the chat, with a chart to prove it. Currently at v7.5, being prepared for white-label rollout to other funds.

1.3Mparticipants covered
9knowledge objectives tracked
3data sources on autopilot
v7.5shipped and running

Let's talk.

Tell us what you're building. We'll tell you whether we can help.