Why the People team should own the AI transformation

Hi Reader!

What do you think is the problem every startup has? Ideas, knowledge, and important questions live in people’s heads and are scattered across Slack, meetings, documents, and email. When someone leaves, it goes with them.
This week on Built by People Leaders, I talked with Stacey Richey, Global VP of People at Smartcat, about redesigning how the work happens and putting the People team in charge of AI transformation.
“Every time somebody left the company… we lost that part of the institutional intelligence.”

Here’s how Smartcat rebuilt the work so that knowledge stays and AI has real context. They call it the intelligence fabric layer.

Record everything. Capture every meeting, call, Slack message, and email into one place. You can’t redesign work you can’t see — recording turns scattered, invisible expertise into signal you can read. “We recorded every single meeting, every single phone call, every single Slack or email.”

Find the one workflow worth automating first. Read the signal for where human work slows the company down or produces bad calls, and pick the single workflow where AI would pay off most. Sales research and deck-building is a strong first candidate. AI spread thin everywhere gets you nothing everywhere; aimed at your highest-leverage work, it changes the economics of the whole function.

Redesign the work around jobs to be done. Reorganize around the missions and problems the business needs handled, not around roles or headcount. Stacey framed it as “jobs to be done and missions to be handled,” and pointed to a Jack Dorsey piece making the same case. Work built around problems, not org charts, is work an agent can actually take on.

Build one source of truth. Make the same information available to everyone, regardless of function, country, or seniority. Teams lose enormous time to alignment and coordination. Shared context removes it. “It’s already there. It’s already built in. We’re already speaking the same language.”

Treat agents as coworkers who learn. Give agents context, memory of past decisions, and feedback — coach them the way you’d coach a new hire. An agent that starts from scratch every time stays junior; one that remembers and improves becomes a teammate. “We don’t talk about AI as tools. We don’t even say AI and tooling in the same sentence. We talk about AI coworkers, AI teammates.”

The People team owns it. AI changes how work gets done, how decisions are made, how knowledge flows, and what skills people need — Stacey’s areas, and probably yours. The function shifts with it: from running processes to designing the operating model, from managing knowledge to designing how intelligence moves, from planning headcount to planning capability — the mix of people, agents, and automation. The leader’s job shifts too, “from having all the answers to asking better questions.”

One caution before you start. If you measure this by hours saved, you’ll call it a win too early and miss the point. Smartcat’s North Star is ARR per full-time employee. But the real test is what the freed hours fund. Stacey’s version: “If AI gives me back 10 hours a week, I don’t consider that a success because I worked fewer hours. I consider it a success because I spent those 10 hours coaching leaders, improving organizational design, solving workforce problems that actually move the business.”

The next episode comes out this Friday - find it on YouTube, Apple Podcasts, or Spotify.

Daria

P.S. On October 8th, I'm running a webinar on the part of AI adoption many teams miss: when the work changes, so does the energy people bring to it — and that decides who thrives and who burns out. Secure your seat here.