The personal-data-vault graveyard: Skiff, Datacoup, Streamr, and the pattern
Every force that ended these applies to us, and by the measures that ended them we are earlier and weaker than any of them ever were. No revenue.
Read this as a record, not as a competitive scoreboard. Nothing here is an argument that we will do better.
Skiff: the exit was the ending
A privacy-focused product acquired in February 2024, and shut down. The sunset was originally six months and was extended to twelve after users objected.
The instructive detail is not the acquisition. It is that accounts were not migrated. Users had to export their own data, manually, before a deadline.
For a custody product that is the failure mode that matters. The pitch was that your data was yours; the ending made that true in the most expensive possible way. A commercial success for its founders was a forced migration for everyone who had trusted it.
Datacoup, paying people for their data pays badly
An early personal-data marketplace that ceased operating. The diagnosis in public accounts is economic rather than technical: individual data is worth very little individually, so the payments were too small to motivate anyone, the pool stayed small, and a small pool is not worth buying.
Our own internal note and the public accounts disagree about the year and the manner of its ending. We are not asserting either, and the disagreement is worth more to a reader than a confident date would be: secondhand competitor records rot, including ours.
The lesson survives the uncertainty: the monetization model needs scale it cannot reach without value it cannot offer first. That is the cold-start problem with a price tag on it.
Streamr, declining, and reporting it themselves
Not dead, and the distinction matters. From their own quarterly transparency report: a major exchange delisted the token in February 2026, and active staked node operators fell from 159 to 93 in one quarter, a 41% drop, with the report attributing it to operators becoming unprofitable after the price fell.
Publishing that about yourself is creditable and we say so.
And it is the case that should worry us most, because of what it demonstrates: when infrastructure capacity is paid for in a token, the capacity is a function of a price nobody controls. The engineering did not change. The operators left because the arithmetic changed.
We do not have this problem because we have not built the thing that causes it. That is not a virtue; it is an absence.
The pattern
None of these died of bad engineering.
- One died of a good commercial outcome that was an ending for its users.
- One died because the unit economics of paying individuals for data do not work at small scale.
- One is shrinking because its capacity was denominated in something volatile.
All three are economic failures, and the theory is on another page, this one is the record.
Where we sit against each of them
Against Skiff: they had users and revenue and an acquirer. We have none of the three. Our storage is a standard Solid pod, so the data is portable by protocol, but there is no export button and no account-deletion path, so "portable in principle" is doing real work in that sentence.
Against Datacoup: we are not trying to pay anyone for their data, so their specific economics do not apply. We also have no revenue model at all, which is not the better position.
Against Streamr: they have operators, a token and a published quarterly report.
On every axis these three were measured by, we are further back.
The question a customer should ask every vendor, including us
"What happens to my data when you stop?"
Our honest answer today: the storage is a standard Solid pod, so another server that speaks the protocol can host it, and there is no export button, no account deletion, and no backup story we can point at. The only unconditional hedge is to run the server yourself, which costs nothing and needs nothing from us.
A vendor that cannot answer that question plainly should be assumed not to have an answer.