AI-as-a-tab is a dead end
The bolt-on chatbot ships fast and demos well, then goes unused. A note on why AI belongs in the flow of the work — and what changes when you move it there.
Watch the usage data on a bolt-on AI chatbot for a few months and it tells the same story almost every time: a spike at launch, then a long slide to near-zero. The reflex is to blame the model or the prompt. We think the problem is upstream of both — it's the tab itself.
The tab charges a tax
A separate AI panel asks the user to stop what they're doing, re-describe their context to a blank box, read a generic answer, and carry it back into their actual work by hand. That's a tax on every single use. It's tolerable once, in a demo. It's unbearable as a daily habit, so people quietly stop paying it.
Context is the whole game
When AI sits inside the work — next to the sentence, the row, the file — it can see what the user sees. Its suggestions get specific instead of generic, and the user can judge them in place. That's the difference between a tool that feels like leverage and one that feels like a detour. The model didn't change; its proximity to the work did.
Why teams default to the tab anyway
Because it's easy to build. A chat panel is a weekend; weaving AI into an existing workflow, with all its states and edge cases, is real product work. So teams ship the easy thing, call it an AI feature, and are surprised when it doesn't stick. The easy path and the effective path point in different directions here.
The exception, so we're honest
Some jobs genuinely suit a conversational surface — open exploration, Q&A over a knowledge base, tasks with no fixed home in the product. The mistake isn't using chat; it's defaulting to it for everything because it's convenient to build. Ask where the work happens, and put the intelligence there.
This is one of our core threads — explored more in inline, not a tab and in how trust is built at the moment of doubt.
Wondering whether your AI belongs inline or in a tab? The free AI Experience & Trust Audit gives you a specific read.
More research
The trust gap: why users abandon accurate AI
A pattern we keep seeing: AI features get abandoned not because they're wrong, but because users can't tell when they're right. Notes on the trust gap and how design closes it.
Read →Research note · 6 minProduct judgment is the new moat
As models commoditize, the defensible edge moves from the technology to the decisions around it. Notes on why product judgment — not the model — is becoming the moat.
Read →