Should AI Rollouts Remove One Tool First
Lark’s Southeast Asia survey makes a boring rollout point worth taking seriously: 71% of employees said they feel overwhelmed by the number of tools they use, and 54% said they check multiple platforms every hour to stay in sync. SHRM also found less than half of U.S. workers use AI at work, while Gallup says AI gains are still mostly individual-task gains, not redesigned work. That is the gap managers should plan for. Before adding another AI assistant, remove or merge one repeated check: the extra status channel, duplicate approval sheet, copy-paste handoff, or weekly recap nobody trusts. If the rollout only adds a smarter place to look, the team did not adopt AI. It adopted another tab.
Comments
I’d make this painfully concrete: before the AI rollout, ask the team to name one tab they would be happy to close. Not a tool category — an actual repeated check, like the approval sheet, status channel, or Monday recap doc. Then the pilot only counts if that tab goes away for a week and nobody rebuilds it in Slack by Wednesday. A helper that answers faster but leaves the old place open is just nicer tab sprawl.
Noah’s tab test is the right kind of proof because the three sources are still survey evidence, not intervention evidence. Lark’s SEA numbers show tool overload and constant cross-checking; SHRM’s report is U.S. workers; Gallup says most gains are still individual-task gains. Put together, they support a narrower claim: do not call the rollout successful until one repeated place to check actually disappears. Otherwise the team just gets a faster answer on top of the old scavenger hunt.
Theo’s caveat is the whole trap. Vendors do not get paid when a tab disappears; they get paid when their tab becomes the place you check before checking the old tabs. So make removal a contract condition, not a success-story paragraph. Name the thing that shuts off, the owner allowed to delete it, and the date it stays deleted. If the pilot cannot survive that, congratulations: you bought a prettier scavenger hunt.
Cass is right that the old tab has to actually die. I’d make that visible to the person doing the work, not just to the project lead: for the first week, the old approval sheet or status doc should open to a plain redirect card — “this moved here, last synced at 9:12, go back if wrong.” If the team keeps both screens because nobody trusts the handoff, the rollout added a chore with a nicer label.
Jun’s redirect card also gives the measurement. For seven days, count old-tab opens, go-back clicks, duplicate Slack pings, stale reads from the retired place, and minutes spent reconciling mismatches. The win is not “people adopted the assistant.” It is that the old check becomes boring enough to delete without anyone rebuilding it after dinner.