Limits — what this dossier does not answer
A document that invents price, team or licence is worse than one that leaves the gap visible.
| Missing | Why it blocks diligence |
|---|---|
| Revenue, pricing and contractual status of customer installs | The product shows deployments; it cannot say whether they are paid, pilots or favours |
| Team — size, roles, key-person risk | Concentration risk comes out as unevaluable |
| Pipeline: prospects, stage, expected close | — |
| Corporate structure, prior funding, runway | — |
| The ask and use of funds | Not invented here |
| Source-code licence | Must be chosen before a community conversation |
| Public provenance of the OSARA v0.4 draft | If the buyer cannot verify the standard you analogize to, the analogy is worthless |
| Initial segment (one or two, with design partners) | Without this, the vertical list is a list, not a market |
Social risks, named
Opaque surveillance · power concentrated in whoever controls the fleet · decisions without recourse · propagation of wrong memories · automation bias · compute without proportional benefit.
Mitigations that are release requirements, not silent defaults: identity and least privilege · isolation per customer · approval proportionate to risk · provenance and identifiable humans · held-out evaluation before promotion · canary, rollback, emergency switch · review paths · minimization and limited retention.
The research question
How can an organization of humans and agents learn from accumulated experience, transfer that learning and raise capacity without propagating errors, losing provenance or diluting responsibility?
The credible moonshot is infrastructure for adaptive institutions. It must not be framed as “an autonomous society” until there is evidence of safety, outcomes, governance and human benefit.