Newsletter review cut from 16 hours to one

CAOP, the organization behind Platformnaarwerk.nl, replaced manual newsletter review with an AI agent that now scans ten times more newsletters in a fraction of the time.

Client

Industry

Public Sector

CAOP is the organization behind Platformnaarwerk.nl, a public sector platform that lists tools relevant to the labor market. To keep the platform current, CAOP follows a steady stream of newsletters about work and employment. We worked with the team responsible for the platform's content.

16 → 1 hour

Weekly time spent on newsletter review

90%+

Of manual review tasks automated

1,000

Newsletters analyzed per week, up from 100

Challenge

Too many newsletters, too little time

Every week, CAOP staff read more than 100 newsletters by hand. They searched each one for new tools and checked whether those tools were already listed on Platformnaarwerk.nl. This took about 16 hours a week. The process relied on the knowledge of individual employees, so their expertise was hard to capture or pass on. There was little structured quality control. As a result, insights that mattered for the labor market reached the platform later than they could have. The workload also set a hard limit: reviewing more sources would simply have meant more manual hours.

Solution

An AI agent that reads newsletters first

We built an AI agent that reads incoming newsletters and pulls out the tools that matter for Platformnaarwerk.nl. It checks each tool against the platform, so the team only reviews what is new. Monthly reports show which tools and trends keep coming up.

Newsletter scraper: collects incoming newsletters for analysis.
AI extraction agent: finds the relevant tools in each newsletter.
Platform check: flags whether a tool is already listed on Platformnaarwerk.nl.
Editorial review: lets editors check and approve the agent's findings.
Monthly analyses: report on trends and how often new tools appear.

Human-in-the-loop

Editors stay in the loop, and their decisions are translated into the system's logic.

Approach

From manual review to AI-assisted editing

Automating the search work while keeping editors in control of what goes online.

Process analysis We mapped the manual review process and pinpointed the bottlenecks: limited quality control and little knowledge retention.
AI agent and scraper We built an agent that extracts relevant tools from newsletters and checks them against the platform.
Human-in-the-loop We set up an editorial review step and turned the team's expertise into rules the system follows.
Monthly analyses We added automated reports on trends and the frequency of new tools.

Built by

Noah van PottenAI Engineer

Jonathan van WestendorpLead AI Engineer

Impact

Ten times the coverage, far less effort

The weekly review that once took two full working days now takes about an hour. Most of the manual work is automated, so the team spends its time on judgment instead of searching. The system handles ten times the previous volume of newsletters. Employee expertise is now part of a process that is scalable and repeatable, instead of living with a few people. Clearer insight into trends strengthens the platform's content strategy. Platformnaarwerk.nl can now respond faster and more consistently to new developments in the labor market.

Contact