Two connected programmes
Viewyonder works beneath the interface, where models meet infrastructure, identity, tools, cost, and consequence.
The work
We follow what an AI application can do, and examine the inference choices that shape what it becomes.
AI application security
The application is the attack surface.
Surfaces is Viewyonder’s programme for investigating the whole AI application: identities, data, permissions, tools, services, and downstream effects—not just the model response.
Injectionator is the working platform inside Surfaces. It exercises meaningful paths, finds the owning controls, and connects an apparent AI outcome to evidence from the systems behind it.
Evidence-backed decisions
The model is an infrastructure decision.
Inference examines the providers, models, routes, regions, plans, contracts, and configurations behind an AI capability. The aim is practical understanding, not a universal leaderboard.
Claims stay attached to their evidence, applicable scope, confidence, and observation date—so a team can see what is known, what remains uncertain, and what the choice means in practice.
Signals from the field
A bounded synthetic example of why model output, application enforcement, and downstream effect must be tested separately.
Why we are putting a customer-proxy agent inside pull requests, and why passing tests is no longer enough proof.
We borrowed /handoff from Matt Pocock, added /pickup, and now we have a paired Claude Code skill that gives us a deliberate, throwaway context bridge across /clear — instead of the usual cargo-cult HANDOFF.md that nobody updates and nobody trusts.
Start with one application surface or one inference decision. We’ll establish the boundary, the evidence, and the first useful question.