Dropbox has completed a leadership handover, with Ashraf Alkarmi taking sole charge and founder Drew Houston moving to executive chairman.
The new chief executive inherits a company that spent two years with essentially flat revenue before posting a single quarter of growth.
His answer is a bet that most of the industry is not making.
Dropbox will not build its own models.
It will not tune or optimise anyone else's.
Instead it argues that the constraint on useful artificial intelligence in the workplace is not model capability but context, meaning the accumulated files, versions and decisions that describe what a project actually is.
That is a self-serving argument from a storage company.
It is also broadly correct.
The reason a general-purpose assistant struggles with a real corporate task is rarely that it cannot reason.
It is that it has no idea which of four documents is current, who signed off on the third one, or why the second was abandoned.
Dropbox has folded its search tool, Dash, directly into its core products so that users can edit, share and change files without leaving the workflow.
The design principle is that a customer's context travels with them across whichever model they choose, open-weight or closed.
The company's job, on this reading, is the unglamorous layer of security, privacy, governance and transparency that sits underneath.
The strategy is coherent, and it neatly avoids competing with far better capitalised laboratories.
The commercial decision attached to it is riskier.
Dropbox is deliberately not charging for Dash yet.
Management says it wants to prove the product can compress a day of work into ten minutes before it puts a price on that, with monetisation models being tested through the coming quarter.
There is a respectable version of this argument, in which premature pricing suppresses the adoption needed to demonstrate value.
There is also a less flattering version, in which a company with weak growth is postponing the moment it discovers customers will not pay.
The three-step plan set out by management, proving growth, proving it is durable, then unlocking more of it, is honest about how early the process is.
Only the first step has any evidence behind it, and that evidence amounts to one quarter.
The product roadmap suggests where the second and third steps are meant to come from.
Dropbox is targeting marketing, creative and construction teams, where feedback still arrives as email and has to be manually translated into file changes.
Replay, a review and collaboration tool, is being rebuilt and relaunched to serve them.
These are sensible, narrow, defensible markets.
They are also small ones, and a company of this size needs Dash to become a paid product eventually.
The patience is admirable, but it has a deadline.
