Heavy LiftersAUG 13, 20262 min read
Heavy Lifters: the people putting AI to work
One usage metric a month, the people leading it in alphabetical order, and one company-wide number worth watching. This month: interactions across Claude and ChatGPT in the last 30 days.
Keenan ChiassonAUG 13, 2026
We believe that to be the best at your job, and for Bigge to be the best at what it does, you have to put these tools to work. So every month this section takes one usage metric across Bigge's AI tools, Claude and ChatGPT both, and names the people leading it.
The list is alphabetical on purpose. This is recognition, not a race, and the metric rotates every month so different kinds of usage get their turn: total messages, tool and skill usage, projects, and the jump from not using the tools at all to using them every day. The data comes from the Innovation team's usage reporting, pulled August 17.
This month: interactions across Claude and ChatGPT, last 30 days
An interaction is one message sent to Claude or ChatGPT, or one Claude Code session run. These ten led the combined count over the last 30 days, and the scale is worth seeing: the top of the board is past 2,400 interactions and the tenth name still clears 500. Not all of it looks the same from the outside, either. Most of the board is chat, while Garrett St. John's month is 1,023 Claude Code sessions running work in the background.
- Brian Noga
- Eddie Gray
- Garrett St. John
- James Iorio
- Jeffery Parrett
- John Leventini
- Jonathan Mast
- Lisa Recuero
- Thom Bostrom
- Will Fike
A running total across all 149 people who used either tool in the last 30 days, busiest first and unnamed. The first ten account for 40 percent of every interaction, half of everything comes from the first 18 people, and it takes 79 of them to reach ninety percent.
A running total across everyone who used either tool, busiest first and unnamed on purpose. Read it as: the ten above account for 40 percent of everything, half of it comes from 18 people, and it takes 79 of them to reach ninety percent. The curve keeps climbing to 149, which is the part worth watching. Plotted in steps of five people.
The number under the number
A monthly total can hide a trend, so it is worth watching the weeks. ChatGPT's export is the only weekly series either platform gives us, and across fifteen complete weeks it is remarkably level: 62,207 messages, averaging 4,147 a week, low of 3,371 over the July 4 week and a peak of 4,739 three weeks later. Whatever else is happening, the work being sent to these tools is steady.
ChatGPT messages for each of the 15 complete weeks from May 4 to August 16: 62,207 in total, averaging 4,147 a week, with a low of 3,371 at the start of July, a peak of 4,739 in the week of July 20, and 4,500 in the most recent week.
ChatGPT messages a week, May 4 to August 16. Claude reports rolling 30, 60 and 90 day windows rather than weeks, so it cannot be drawn on this axis.
Underneath that flat line is the number this department actually watches. Seats grew from 131 in May to 145 in late July, and the count of people using one in a given week did not: it drifted from 116 down to 102. Unopened seats doubled, from 15 to 31, and the share of seats in use fell from 89 percent to 77. Steady volume from a shrinking group of people is not the same thing as adoption, and closing that gap is the job.
ChatGPT seats each week, split into the ones used and the ones nobody opened. Seats rose from 131 in early May to a peak of 145 in late July before settling at 133, while the people using them fell from 116 to 102. Unopened seats went from 15 to 31, and the share in use from 89 percent to 77 percent.
Every ChatGPT seat, week by week, stacked into the ones somebody used and the ones nobody opened. The hatched band is the gap. Press a figure above the chart to isolate either half.
Not on the board yet?
Uncomfortable with AI? Have questions? Not sure where to start? That is exactly what the Innovation team and this department are here for. Ask in the AI General channel on Teams, and someone will get you moving. Next month's metric is a different one, and the board resets every issue.
