Shovels Shovels - Building permit and contractor intelligence
Why AI has us hiring more people, not fewer—and where they're going

Monthly Memo - November 2026

Company
Ryan Buckley

Ryan Buckley

CEO


A couple of months ago, I wrote about AI, taste, and credit cards. This month, I'm writing about hiring, because we've been doing a lot of it. We added six teammates in the last two months, and we'll probably double our headcount in 2027 — because of AI, not in spite of it.

Here's why.

Storm: data you can't Google

Our Storm team brought 364 new jurisdictions into the Shovels production data release this month. Another 600 are already sending us data that still needs parsing — the harder stuff, messy PDFs and the like.

These cities and counties don't publish their records online, so don't bother asking Claude or Google for this data. We launched Storm more than a year ago to go get it. The team finds the right person in a city or county, works through the public records request process, and gets data flowing to Shovels. We call that activation. It's human-to-human work: understanding how a jurisdiction operates, answering questions, and earning trust. That part doesn't automate.

Once the data is flowing, agents handle the recurring requests and follow-ups. The Storm team manages those agents, handles the exceptions, and stays accountable to the people sending us data.

That's the pattern. The more routine work the agents take on, the more time our people have for the work only people can do. So next year we're making a major investment in Storm, with more people and better tools, to 10X our reach and get more than just building permits.

Engineering: harnesses, not just code

Getting a jurisdiction to send a file is only the beginning. We still have to process it, clean it up, and make it useful in Shovels. That's real engineering work, so we're hiring an AI ops and data engineer (probably several, tbh) to work directly with Storm on its agent platform and data pipeline.

Across engineering, we plan to hire one to two engineers every month in 2027. AI makes it faster to write code, but it also makes it easier to pile up technical debt. More code isn't the goal. Reliable products and useful data are. Our engineers build the harnesses around our agents: the tests, checks, and reviews that let them do useful work without creating a bigger cleanup job later. Then they take responsibility for what ships.

More to do than ever

The other reason we're hiring: we're loaded to the gills. We had a banger of a Shovels Summit that reminded us what's possible and how much is still in front of us. A few examples:

  • Google Sheets plugin (already live!)
  • Nationwide contractor IDs (up next)
  • Nationwide project tracking (also up next)
  • Decisions in Shovels Online (already live!)
  • Team Management in Shovels Online (Q4, maybe Q1)
  • Hundreds of once-offline jurisdictions (already live!)

We can do all of this at once because we finally have the people-power. It took three years to earn it. Now we're going for it.

If you want to join the party, we're hiring, obviously: shovels.ai/careers