Evidence / Agenda / Shared language

A working foundation.
An exact account.

Inspect the evidence, examine the open questions and identify where the next application matters.

What we think we have solved

Automation cannot authorize its own protected consequence.

ZLAR is a working lock for human authority. We believe that separating capability from the permission to produce a protected effect establishes a foundation for safe automation. The next task is to make that foundation useful across more destinations and workflows.

Dated implementation evidence

Two implementations.
Separate records.

Bramah / Sepolia / 29 October 2026

An approved transfer.

Public Bramah test-wallet records connect approval, submission and execution. The matching test-network transaction records a successful transfer.

Public event records
Matching transaction

Separate wallet demonstration / 4 November 2026

Approval, refusal and expiry.

ZLAR’s review records report an exact payment approved on a phone, a refused request without a transfer, and an unanswered request expiring without a transfer.

Approved payment transaction

The public transaction establishes that transfer. The phone interaction and other cases rely on the separate review observations.

Reading the evidence

These are dated results from distinct implementations. They do not establish every integration, attack class or current installation. A permission mechanism does not by itself prove a permitted action is correct or beneficial. The records describe the implementations and dates shown.

The November wallet observations do not establish stolen-server-key resistance. Preserve that implementation’s scope when considering its results.

Access to the Bramah Challenge is by invitation. Ask about an invitation

What we are working on

Make the boundary useful in the world.

01

Destination adoption

Identify the consequence, the accepting paths and the integration that makes human authority binding there.

02

Authority across composition

Preserve scope and allowances as work is delegated, run concurrently or assembled into a larger result.

03

Recovery and changing conditions

Keep uncertain outcomes, changed proposals and withdrawal distinct. Make the next supported step useful.

04

Authority over authority

Examine who can change the rules, approvers and protected boundary. Those changes have consequences too.

05

Human value and adoption

Make decisions understandable, remove unnecessary interruption and demonstrate value in a real workflow.

This is the product and research agenda. Individual integrations need their own implementation and evidence.

Shared concepts / G01–G15

Language precise enough to build with.

G01Consequence
An effect of work: a change, disclosure, resource use, commitment or delayed effect.
G02Accountable human authority
Relevant human decision power and responsibility for a consequence, with scope and limits. Technical access alone does not establish it.
G03Standing permission
A bounded allowance that lets work continue while its scope and conditions remain valid.
G04Exact yes
Approval for an identifiable change, with its material context, limits and uncertainty made understandable.
G05Grant
A record of conferred authority under an applicable recognition arrangement.
G06Credential / key
A credential presents evidence for a check; a key signs or authenticates. Its significance depends on the authority and scope recognized for it.
G07Refusal
Permission withheld by a human or a request rejected by a service. Preserve its source and reason; keep execution outcome separate.
G08Withdrawal
A later decision taking back permission. The request, effective stopping and remaining effects are separate events.
G09Receipt
An attributable account of a reported decision or outcome. Its issuer, scope and verification context determine what it establishes.
G10Uncertainty
What remains unestablished, including whether an attempted consequence occurred.
G11Review
Examination by someone with the relevant information, capacity and decision power.
G12Remedy
An authorized practical response to harm, such as correction or containment. An explanation alone is insufficient.
G13Human independence
Practical ability to understand, choose, delegate, question, refuse and obtain help.
G14Protection boundary
The specified destination, effect and accepting paths over which protection applies.
G15Benefit and distribution
What improves, who can access it, who bears costs or disruption, and how control changes.

About this edition

A public thesis, open for review.

ZLAR is the technology. ZLAR Inc. is the company. This edition of ZLAR.ai, dated 6 November 2026, is for founders, AI insiders and the AI they work with.

The diagrams explain distinctions; the deployment states are fictional examples. Application families describe the opportunity, not a catalogue of installed integrations. The evidence section identifies dated records separately.

This site contains no submission form, account connection or application storage. It counts visits in aggregate with Cloudflare Web Analytics, which uses no cookies and builds no visitor profiles. Email links open the visitor’s own email application, and what you send is used to read and answer your message.

A general description is enough to begin a conversation. Use appropriate authorized channels for confidential information.

Build ZLAR with us

Bring a difficult consequence.
Bring the expertise to make it useful.

Architecture, product, commercial judgment or a consequential workflow. Start a conversation about turning this foundation into widely used infrastructure.

Talk about building ZLAR