14% of Workers Have AI Brain Fry. One Inbox Fixes It.

Fourteen percent is not a rounding error. In a study of 1,488 U.S. workers, Boston Consulting Group and UC Riverside researchers identified a cluster of people experiencing what they called AI brain fry: mental fog, trouble focusing, headaches—the feeling that the tools meant to lighten the load are instead draining the tank. Those workers were 11% more likely to make minor mistakes and 39% more likely to make major ones. They were also more likely to be eyeing the door. Marketing teams showed the highest prevalence, at 26%. Coverage of the findings appeared widely, including in the Indian Express. The story is not that AI is useless. It is that the way we are using it—sprawled across tabs, apps, and dashboards—is expensive in a currency we rarely measure: sustained attention.

Separate research sharpens the picture. The Human Clarity Institute reported in 2025 that 43% of professionals say checking AI outputs for accuracy pulls focus away from other work, and more than half of those find prompt refinement mentally taxing. A piece in Frontiers in Education ties similar dynamics to cognitive fatigue in knowledge work. Related commentary on workload intensification from AI tools appears in AAPL. The headline writes itself in a grim way: we asked machines to think faster, and humans got tired faster instead.

What "brain fry" actually looks like on a Tuesday

Picture the loop. You draft in one window, fact-check in another, summarize in a third, and paste everything back into email so your team can see it. Each hop is small. The tax is not. You are not only switching software—you are switching mental models, tone, and what "done" means for each tool. That is the opposite of how email has always worked: one thread, one place, one protocol everyone already understands.

The via.email bet is that specialization does not have to mean more interfaces. You keep living in the inbox. Each task goes to an agent with a narrow job and a clear address. You are not learning a new operating system every quarter.

When more tools stop helping

If you have read our earlier piece on why one interface beats a dozen tools, you know the through-line: cognitive load scales with context switches, not just with task count. The new BCG-linked numbers add a performance angle—errors and attrition intent—that managers can actually cite in a budget meeting.

The pattern also shows up where productivity peaks and then cracks after a handful of AI tools—not because model quality falls off a cliff, but because orchestration becomes its own full-time job. Throw in the copy-paste tax between apps and the hundreds of billions tied to context switching, and the case for consolidation stops sounding like vendor poetry and starts sounding like arithmetic.

Email as the narrow front door

Here is the move that sounds almost too simple to work: stop treating email as the graveyard where finished work gets filed, and start treating it as the place where work gets handed off.

Forward a dense thread to Timeline Threads timeline.threads@via.email when you need a clean chronology before a hard conversation. Forward a newsletter dump to Extract Newsletter Insights extract.newsletter.insights@via.email when you want signal, not slogans. When a chain has turned into a pile of implied tasks, Extract Action Items extract.action.items@via.email pulls owners and deadlines into a list you can actually execute. If you only have ninety seconds, Distill to Three distill.to.three@via.email forces the three bullets that matter.

None of that requires a new login, a Chrome extension, or a training webinar. It requires the thing your company already standardized twenty years ago: an email address.

What via.email can and cannot do for you

Agents run on the text and files you send in-thread, and they write back in the same conversation. That is how via.email keeps the surface area small. It also means the constraints are honest: there is no persistent memory across unrelated threads, no access to your inbox or calendar, no autonomous sending on your behalf, and no long-running monitoring. You stay in control of what gets forwarded and when. For the full capability list, see pricing and tiers; heavier inputs like attachments and live web search depend on your plan.

The takeaway

AI brain fry is not a moral failure of employees who "cannot keep up." It is a systems problem: too many surfaces asking for judgment calls per hour. Shrinking the number of places where those calls happen is not laziness—it is risk management. One protocol, many specialists, zero new dashboards to babysit. That is the fix hiding in plain sight, in the app you already have open.