U.S. payroll rules change constantly, and the updates rarely arrive in one place. So I built one. It is a Notion page tracking minimum wage changes, payroll compliance updates, and final pay laws across states and localities, and every item carries the effective date, a summary, the source, and any notes. A set of AI agents rescans for changes every week and refreshes the page, which shows when it last updated. Before this, a minimum wage change came through one source, a final pay rule through another, and a compliance note through Slack.
I'm Nick Halfen, Senior Payroll Specialist Lead for the U.S. at Remote. I had just completed my second month here when I built this, as part of Remote's internal AI training, now available publicly as AI for Actual Work. I did not have to look far for a project. The information gap was in front of me every week.
Our U.S. payroll team is growing consistently as more customers choose Remote, which means more questions and more edge cases to keep straight. We need a reliable way to track payroll compliance, minimum wage, and final pay laws, and the information cannot stop with our team. The customer experience team has been asking for it lately, because customers need answers. Payroll changes also have to reach the resources customers rely on. Some of this used to come through Slack, including Watchtower payroll updates, our internal alerts for payroll related regulatory changes. The rest arrived piecemeal. What I wanted was one place where anyone could see the current picture without assembling it themselves.
What the tracker covers
The first version was a Google Sheet, a shared tracker pulling the updates from the past four weeks and sorting them by category: minimum wage in one section, compliance in another, final pay in a third. Each row carries the state or locality, the effective date, a summary, the source, and notes. The source column is not optional. Payroll work needs a way to go back and verify where something came from, and there should always be a backup for the backup.
Final pay needed a little more than a row. Some states require a much faster final payment timeline than others, so I added red formatting to make those stand out. As payroll employees, we can then see at a glance which situations move fastest.
What broke along the way
Then I set up an architecture with multiple agents to search for current U.S. payroll compliance updates, minimum wage changes, and final pay laws, with a baseline for 2026 and instructions to keep capturing changes going forward. The weekly refresh became the maintenance rhythm, and it was also the first thing to break. The workflow could only update a local Google Sheet, not the master sheet I intended to share, so I connected Claude to Chrome and that solved it.
Then I thought the spreadsheet was the wrong destination anyway. A Notion page made more sense for something I wanted to share across the company, so I asked Claude to move everything from the sheet onto a page. It opens with the snapshot, then minimum wage, compliance, and final pay. The top shows when it last updated, and at the bottom I documented the framework reference: the weekly compliance scan, the sources it uses, and how to reach me if something needs a second look.
One thing I probably should have done from the start is send everything straight to the Notion page. Updating both the Google Sheet and the page occupies more usage than it needs to, so reducing that duplication is the next tweak. I am also monitoring the scan errors and the update errors each week to see where the workflow needs attention.
From there I can see this going in two directions. The same setup could support similar pages for individual countries, giving payroll teams a trusted reference in Notion that updates itself instead of something they rebuild by hand. It could also connect more directly to the customer facing resources that depend on payroll data, so a change we catch this week reaches customers much sooner.
Payroll judgment has not gone anywhere. Somebody still has to check the sources and decide what is solid enough to put in front of a customer. Some of this information will also only change infrequently, so what matters is knowing it is current when it does change, with the source right there to verify. There is a lot of potential in the approach, and I am looking forward to seeing how far it goes.
AI for Actual Work is the free course we built from Remote’s own training. Sign up at aiforactualwork.com to start building with AI in your day-to-day work.
