Partners
Built with others, not alone
Parts of NAOMS come from people outside the core team who did that particular piece better. That door stays open on purpose — every building block in NAOMS can be replaced by a stronger implementation, and the team names the piece it thinks is weakest first.
By David Thomson: versioned, forkable AI orchestration recipes, with their provenance anchored on-chain so a recipe's history is provable, not just documented. Integration work is underway now.
Aqua Protocol, by inblockio: the tamper-proof pack format NAOMS uses for everything that leaves a device — backups, shared memories, and installable packages. A pack carries proof of where it came from and that nothing inside it changed since, so a corrupted or interfered-with pack is caught rather than silently trusted.
By Lumina Nao: a disk-backed slot cache for local model inference. Idle coding sessions park their context on disk instead of holding it all in memory at once, so one GPU machine can run several sessions in parallel with contexts larger than it could otherwise keep in RAM together. NAOMS has run its own real-model serving through this pairing since June 2026.
recrypt, by identikey: proxy re-encryption. It lets someone grant another person access to their encrypted data later, without ever handing over their own key or decrypting the data in the middle.
Verification by people, not a company. When someone wants a claim on their Relationship Card checked, such as an email address or phone number, a randomly chosen panel of hive members verifies it and co-signs the result with a threshold signature. It includes dispute handling, verifier reputation and age checks that reveal only what is needed.
The open standards NAOMS builds identity on, developed with Linux Foundation Decentralized Trust, Trust Over IP, the Decentralized Identity Foundation and the OpenWallet Foundation: Relationship Cards, verifiable relationship credentials, hives as verifiable communities, and personhood.
From Trust Over IP: a standard way to describe the tasks trust systems perform for each other. NAOMS is expressing its APIs as trust tasks, so other trust systems can call them in a shared language.
Build a better block
Every building block in NAOMS is open to being replaced by a better implementation from outside the project. The team's own example of a block they'd like to see improved is agent memory — if you think you can do that piece, or any piece, better than what's there today, that's a standing invitation, not a hypothetical one.