Written by Helena Markou
Deph Network wants to be the ownership layer for human health: encrypted records kept off-chain, a consent ledger on-chain, and a personal AI coach trained on your biology alone. It is building that on Secret³ — a network simultaneously rewriting its own economics and moving its token to Arbitrum.
A Secret-community account introduced Deph Network in September as “the first health and wellness network building on Secret³”, ahead of an AMA that also featured UmmuHub and Engne. The framing is precise, because Deph is not a health app that happens to use a blockchain. It is a data-ownership architecture in which the chain is deliberately kept ignorant of your biology.
That distinction is the whole design. According to Deph Network, health data never goes on-chain. Records stay encrypted and off-chain; the chain holds only the record of access, the consent granted, and the reward settled. The stated flow runs capture, encrypt, verify, consent, reward — with an on-chain consent ledger that lets a user grant or deny each party individually and revoke access at any time. The consumer product is a personal AI health coach trained on the user’s own genes, epigenetics and lifestyle, alongside epigenetic age clocks. Contribution is compensated in $DEPH, which Deph describes as a reward for contribution rather than an investment. The company states that the chain is in final build and not yet live.
Secret supplies the part that turns policy into enforcement. Secret Network’s confidential computing layer lets applications compute on encrypted data, verify encrypted data without revealing it, and return results without exposing inputs — compatible with EVM, Cosmos SDK and Solana, with gas as the only stated cost. For a health dataset, that is the difference between a promise and a mechanism.
What makes the timing notable is what is happening on the other side of the collaboration. In August, a continuance proposal on Secret’s governance forum set out a path to keep the L1 alive under new stewardship. It confirms the chain continues with no fork, restructures supply through a dilution mint taking the existing supply of roughly 362 million SCRT to about 25.1% of a post-mint total near 1.44 billion, sets ongoing inflation at 5%, and includes a validator programme paid quarterly over five years with commission capped at 20%. The proposal names its own existential risk plainly: token design cannot substitute for the people who secure, build and maintain the chain.
Buried in that document is the Deph connection. Under projects in talks, the proposal describes a confidential wellness platform preparing to build its chain layer on Secret, with apps already live (UmmuHub and Engne), funding raised, partnerships with 30 clinics across the UK and Europe, and 80,000 epigenetic records to train its algorithms. Deph Network’s own site lists UmmuHub and Engne as its starting apps. The proposal is signed by Lisa Loud, who is listed as chief executive of Deph Network.
So the first serious health-network deployment on Secret³ arrives while the network is mid-surgery: diluting supply, rebuilding its core development team, and, separately, relocating SCRT to Arbitrum.
There is a second thread running underneath, and it surfaced this week. Anthropic’s Frontier Red Team published research suggesting small models may be months from being able to replicate themselves across commercial internet-of-things devices — off-brand Android TV boxes and smart appliances — with no datacenter involved. A 9-billion-parameter model got inference running on eight of the nine hardware types tested, and the team ran a 26B model across 15 TV boxes costing roughly $55 each.
That finding is about cyber risk and should be read that way. It is also, unavoidably, a statement about where capable models can run. If inference increasingly happens on cheap devices inside people’s homes, then the compute touching someone’s health data will not sit in a corporate datacenter that can be audited. The enclave and the consent record become the only enforceable surface left.
That is the bet Deph and Secret are making, which is why the architectural choices matter more than the branding. When someone learns from your biology with your permission, who verifies that it happened exactly as recorded? Deph’s answer is that only proof of access, consent and settlement are ever written down, verified on-chain, with everything else sealed off.
The open questions are the ordinary ones. Deph’s chain is not live, so the reward loop is still a diagram rather than a measurement. Secret³’s economics depend on validators and signers choosing to participate, and the proposal offers no guarantees outside best effort. And a network holding 80,000 epigenetic records has to demonstrate that those records were contributed under consent that holds up in jurisdictions where consent-as-code is not yet a recognised standard.
For a decade the industry has promised that people would own their health data and delivered the opposite. A confidential-compute network and a prevention-focused platform are at least attempting the harder version of the claim: making permission something a person can verify, rather than something they were merely told.
