A design philosophy
Don't ask users to trust software.
Give them reasons to.
TrustWare™ is software designed so that users can understand what is known, what is uncertain, where information came from, what may be missing, and why a conclusion was reached — before they are asked to trust or act on it.
Plausibility is no longer scarce. Belief is.
Generative AI makes plausible text, images, analysis, and recommendations nearly frictionless to produce at enormous scale. When generation is cheap, plausibility can no longer be the standard for truth — the interface has to expose the basis for belief. TrustWare should never mean “you can trust whatever the software tells you.” That would be exactly the wrong promise. Trust has to be earned — through transparency, evidence, provenance, uncertainty, explainability, and human agency — not asserted.
The gap TrustWare is built for
AI has made generating intelligence cheap. It has not made evaluating that intelligence cheap. As models grow more capable, the tools for assessing what they get wrong, tracing what they relied on, and holding them accountable are not keeping pace — incident reports are rising, transparency among major model developers has been declining, and even leading models can be steered off course simply by how confidently a false premise is stated to them.
AI solved the problem of generating intelligence. It did not solve the problem of knowing when that intelligence deserves our trust.
TrustWare is designed for that gap.
Belief is not evidence.
A person might say “I was told this isn't enforceable,” or “I think this only applies if I terminate early.” A system that treats that premise as fact will reason from a false foundation without ever flagging it. TrustWare keeps these apart on purpose:
Collapsing these into one confident-sounding answer is how plausible systems produce wrong conclusions. Keeping them separate is not a UX preference — it's an architectural requirement.
A new design era
From provenance of content to provenance of understanding.
The Seven Principles of TrustWare
Truth over persuasion.
The system's job is not to convince the user. Its job is to help the user understand.
Evidence before assertion.
Important claims should be connected to their supporting evidence.
Provenance by design.
Users should be able to understand where important information originated and how it changed.
Uncertainty made visible.
Software should distinguish what is known from what is inferred, disputed, incomplete, or unknown.
Blind spots surfaced.
The system should not merely answer the questions users ask. It should help reveal important questions they may not know to ask.
Reasoning made traceable.
Users should be able to reconstruct what evidence, assumptions, and considerations led to an important conclusion or commitment.
Humans retain judgment.
The system informs, challenges, and clarifies. The human ultimately decides.
The TrustWare Architecture
The Seven Principles describe what TrustWare commits to. These six layers are how that commitment gets built — each one answering a different question.
Evidence Layer
What supports this conclusion?
Provenance Layer
Where did the information come from?
Truth-State Layer
Is this fact, stated belief, inference, estimate, conflicting evidence, or unknown?
Uncertainty Layer
What doesn't the system know?
Blind-Spot Layer
What relevant questions or evidence may be missing?
Human Judgment Layer
What does the user ultimately decide?
KnowBefore is the first expression of TrustWare.
The dominant AI interface is a straight line: prompt in, answer out. TrustWare replaces that line with a path — question, evidence, provenance, truth state, uncertainty, blind spots, understanding, and only then, human judgment. KnowBefore never just tells you the answer. It shows you what's known, the evidence behind it, where it came from, what's being assumed, what conflicts, what remains uncertain, and what you may not have considered — then it lets you decide.
Don't ask users to trust software. Give them reasons to.