VERIFIABLE PROOF SYSTEMS
Infrastructure for verifiable authority.
Software now proposes and performs consequential actions faster than anyone can review them.
Verifiable Proof Systems treats a proposed action and an authoritative one as two different things. Whether a proposed effect becomes real is decided separately, against evidence: whether it is authorized, in scope, and backed by a record anyone can re-run and get the same answer.
Intelligence may propose. Authority must be independently established.
Outputs are not proof.
An output can look correct and still leave the important questions unanswered: where the data came from, who authorized it, which rules it followed, and whether any of it can be checked afterward.
Verifiable Proof Systems separates proposing an action from admitting one. A proposed effect becomes authoritative only when the evidence supports it: that it is authorized, within scope, and backed by a record someone else can re-run and get the same answer. Not the system's word for it.
We are early, and we say so. What exists today is present-state assessment: an evidence-based review of what a system is and does. The enforcement layer is designed and documented, not yet runtime-enforced. We publish what is proven and what is not.
For now, a person stays in the loop as the authority within a defined boundary. That is deliberate, and it holds until a shared, verifiable ground between people and machines is actually established, not assumed.
Establish reality. Bound authority. Control consequence.
01
Establish reality
Determine what the system presently contains, implements, demonstrates, contradicts, and leaves unresolved.
02
Bound authority
Separate requested intent from the authority actually granted to a person, model, process, or system.
03
Control consequence
Require evidence and admissibility conditions before consequential state becomes authoritative.
No action becomes authoritative merely because it was requested, generated, signed, or confidently presented.
ASSESSMENT
System Reality Assessment
Before expanding automation, raising capital, entering diligence, or increasing operational consequence, establish what the system actually is.
The Vps System Reality Assessment is a structured, evidence-led examination of a software system's present state. It evaluates artifacts, implementation paths, demonstrated capabilities, contradictions, dependencies, and unresolved conditions.
The assessment does not certify a system, predict its future development, or convert incomplete evidence into confidence. It creates a defensible starting point for technical, operational, and investment decisions.
REQUEST A SYSTEM REALITY ASSESSMENTASSESSMENT OUTPUTS
APPROACH
Evidence before assertion
| COMMON APPROACH | Vps APPROACH |
|---|---|
| Begin with stated capability | Begin with present-state evidence |
| Treat a request as authority | Require authority to be explicit |
| Rely on persuasive output | Separate reasoning from consequence |
| Collapse unknowns into conclusions | Preserve unresolved conditions |
| Review activity afterward | Govern consequential state changes |
| Trust documentation and attestation | Examine artifacts and demonstrated behavior |
Vps is not another trust dashboard.
It is an effort to make authority, evidence, uncertainty, and consequential state mechanically distinguishable.
ENGAGEMENTS
Design partnerships
Vps works with technical founders, small teams, maintainers, and organizations operating consequential software systems.
Suitable partners are building systems where incorrect, unsupported, or unauthorized actions can create material operational, financial, safety, security, or compliance consequences.
DISCUSS A DESIGN PARTNERSHIPEARLY FOCUS AREAS
RESEARCH LINEAGE
Research lineage
Vps is an independent body of work informed by public research in risk management, secure systems, software assurance, formal methods, and software understanding.
SELECTED PUBLIC RESOURCES
- NIST Artificial Intelligence Risk Management Framework (opens in new tab)
- NIST Generative Artificial Intelligence Profile (NIST AI 600-1) (opens in new tab)
- Sandia National Laboratories: The National Need for Software Understanding (opens in new tab)
- Software Understanding for National Security Technical R&D Roadmap (SAND2024-16656R) (opens in new tab)
- Software Understanding for National Security public initiative (opens in new tab)
These resources provide context for the problem space. Their inclusion does not imply endorsement, partnership, certification, or regulatory conformity.
FOUNDER
I came to this problem through operations, audits, compliance, and environments where mistakes had immediate real-world consequences.
Vps did not begin by asking what increasingly intelligent systems could do. It began by asking what no system should be permitted to make consequential without evidence, bounded authority, and an independently enforceable decision boundary.
Adam B. Straughn
Founder, Verifiable Proof Systems