Voice-first human review for AI workflows.
AVRT is being developed as a voice-first human review and authorization layer for AI workflows. Created by Jason Proper, founder of BGBH Threads LLC, it is designed to help people pause, examine a proposed action and keep consequential decisions under human authority.
Development-stage prototype. Private demonstrations and commercial discussions are open.
Built from truth.
Designed for human review.
Our aim is practical: give the person a clear place to review what is proposed, consider consequences and record their decision.
Our Core Values
Human authority
Assessment supports a decision; it does not make a human decision for them.
Clear limits
Show what the prototype did, what it did not do and where uncertainty remains.
Traceable decisions
Preserve a reviewable record without treating a record as proof that its contents are true.
A pause before consequential action.
The intended role is to complement AI workflows with explicit human review and authorization. Connections to external enterprise systems remain planned work.
Fluent output still needs review.
An AI response can sound persuasive while missing context or making a mistake. A proposed action needs a separate decision about whether, when and under whose authority it should proceed.
Review the proposal
What exactly is being requested?
Check the uncertainty
What information is missing or unsupported?
Keep the human decision explicit
An assessment or spoken summary is not permission to execute.
What the combined prototype demonstrates
The current private AVRT Lab development app combines proposal entry, local assessment, fixed spoken summaries and human decision records in one iPhone/iPad prototype.
Founder-run iPhone testing has exercised dictated proposal capture, assessment presentation, matching spoken-summary playback, an operating-system authentication prompt, receipt creation and an in-app chain check.
A note on these observations
These are internal prototype observations, not independent validation or a production-readiness claim.
Propose. Review. Record.
Propose
Enter a synthetic action by typing or dictation.
Review
Read the assessment and listen to its matching fixed summary when available.
Record
Choose a human decision, complete the applicable authentication prompt and inspect the resulting receipt.
The demonstration uses recording-only or explicitly labelled local mock behavior. It does not send email, transfer money or control another company's system.
A working prototype with open questions.
The combined app has demonstrated voice output. Assessment quality, dictation reliability and lifecycle behavior still require work. iPad acceptance is partial.
Independent evaluation, external integration, network-disabled operation and suitability for enterprise deployment are not established.
Advanced Voice Reasoning Technology
A Human-First AI Accountability Framework
A human review and authorization layer in development.
AVRT is intended to support a clear boundary between a proposed action, an advisory assessment and an authorized human decision. Public demonstrations focus on observable behavior; confidential implementation details are not published here.
What we are building toward
Voice-first participation
Make proposals and summaries easier to review through speech as well as text.
Explicit human authority
Keep assessment, authentication, recording and execution distinct.
Accountability
Make decision records inspectable while explaining what their checks do and do not establish.
These are product goals; integration and effectiveness must be evaluated in each intended environment.
AI proposes. A person reviews. The person decides.
A proposed action is brought into the review workflow.
An advisory assessment identifies consequences or missing information.
A person chooses the next step and records a decision.
A spoken summary communicates the displayed state. It does not approve or execute the action.
Voice output and decision recording are present in one development app.
Founder observations and local checks support specific tested behaviors.
Qualified evaluation and a defined commercial development process.
Human authority should remain explicit.
Jason's product direction centers on helping people pause, review and make accountable decisions when working with AI.
Jason Proper
Founder of BGBH Threads LLC and creator of AVRT.
Jason leads the product vision, voice-first requirements and founder demonstrations. He is seeking a team that can carry qualified technical evaluation, implementation and commercialization alongside him.
Commercial discussions should define a continuing, compensated founder role with clear responsibilities.
Building from lived experience.
Documenting every step.
Truth. Honesty. Transparency.
Describe observed behavior accurately. Preserve human authority. Make uncertainty and limits visible.
What a demonstration does—and does not—establish
Assessment is advisory
Model output can be incorrect, incomplete or unavailable.
Authentication has limits
An OS prompt is not independent proof of a named person's identity.
Records have limits
An integrity check does not prove a described event happened.
Integrations need evidence
A local prototype does not establish enforcement in an external system.
Define the test before making the claim.
We welcome a qualified technical team to define useful evaluation criteria, inspect an appropriate protected demonstration and determine what further evidence a potential partner needs.
A private demonstration is an opportunity to assess fit.
It is not a guarantee of performance or a commitment to an engagement.
Human review where it matters.
Our goal is to complement AI capability with a clear, usable moment of human authority. Turning that goal into deployed integrations requires technical diligence and a capable implementation team.
Open to a defined commercial path.
Funded evaluation
Agree a bounded scope, qualified reviewers and meaningful acceptance criteria.
Licensing and co-development
Examine technical fit, required integration work and a protected commercial process.
Strategic investment and acquisition
Explore alignment with a partner capable of developing and scaling the work.
Founder involvement
Define Jason Proper's continuing, compensated role, responsibilities and contribution.
These are opportunities for discussion; no partner, transaction, funding or customer commitment is being claimed.
Start with a private conversation.
For a private demonstration or a discussion of funded evaluation, licensing, co-development, strategic investment or acquisition, contact Jason Proper at BGBH Threads LLC.
These links open your email application. Please keep initial inquiries nonconfidential.
AVRT
A voice-first human review and authorization layer in development.
Jason Proper | BGBH Threads LLC
Contact
© 2026 BGBH Threads LLC. All rights reserved. | AI drafts. Humans review. Humans stay accountable.
AVRT