The request arrives in Slack at 9:22 in the morning. Someone on the support team needs to know the current refund window for annual plan customers. They know the answer exists somewhere. They just do not know where, and so they ask the person who always seems to know things. That person is in the middle of something. The thread sits unread for twenty minutes, then gets a DM, then finally gets answered.
Nobody designed this system. It is what happens when knowledge is written down in one tool and the person asking lives in another.
The knowledge is already written down
The answer to that support question exists in at least three places: a Notion page titled something like 'Customer Policy Notes,' a Confluence article the head of support wrote during the Q2 reorg, and a Slack message from the VP of Operations in November where she clarified the edge case. None of those documents are wrong. None of them agree exactly. And none of them are findable by a person who does not already know they exist.
The knowledge is not missing. It is scattered, across the tools where the decisions that produced it happened. The ticket where an edge case got resolved. The meeting note where the policy was first articulated. The email thread where a customer pushed back and a clearer commitment was made. Those are not searchable as a single body of knowledge. They are only findable if you already know where to look.
Search does not cross tool boundaries
The standard response to this problem is to add search. Every tool has a search bar. Some now have AI-powered search. Teams add them, and the question keeps coming in, because search only works if you know which tool to search and roughly what terms to use. A person who was not present when the policy was written does not know to check Confluence. They find nothing and ping a colleague instead.
The more ambitious version of this is a wiki migration: gather everything into one place, build a taxonomy, keep it updated. This works as long as the migration happens, the taxonomy is maintained, and the knowledge stops accreting in the tools where work actually happens. In practice, none of those three conditions hold for long. People write notes where they are working, not where the knowledge is supposed to live.
The interruption multiplier
There is a less obvious version of this cost. When the same question keeps arriving and the same person keeps answering it, that person starts to be known as the place to ask. The knowledge becomes less distributed, not more. The expert gets more questions, not fewer. The colleague who could have found the answer in a document pings the expert because pinging the expert is faster, and the loop closes.
This is how expertise becomes a bottleneck without anyone intending it. People are rational. They reach for the path of least resistance, and often the path of least resistance runs through a person who knows things rather than through a document that requires knowing where to look.
- The person being interrupted loses an average of 23 minutes of recovery time after a context switch. Multiply that by the colleagues who get pinged throughout the day.
- The person asking waits, sometimes for hours, while their own work sits.
- The answer that eventually arrives is often slightly stale: the colleague answering from memory does not know about the update from two months ago.
- The person who wrote the policy and gets asked about it most often eventually becomes the reason the team cannot move as fast as it should.
The bottleneck in most teams is not a gap in knowledge. It is a gap in access to knowledge.
A question should not require knowing who to ask
Frond connects to wherever your knowledge lives and reads all of it. Ask 'what is the refund window for annual plan customers?' and it retrieves the relevant passages from the Notion page and the Confluence article, reads them together, and returns a clear answer with the exact sentences it drew from, linked to the source documents so you can check them yourself.
The answer is not a summary of your documentation. It is a reading of the exact text where the answer lives, with the source shown, so you can verify it the same way you would if a trusted colleague had looked it up and reported back.
The same questions will keep arriving. The only variable is who pays the cost of answering them each time, and whether the answer they get is the one that is actually current.