AXES.
The open standard for automation execution evidence. One execution. One evidence record. Many professional truths.
Magentix AI stewards AXES in the open. The standard evidences; it never certifies.
What AXES is for.
Wherever a consequential action runs without a human at every step - a payment, a manufacturing step, a standard operating procedure, a supply-chain hand-off, a customer or HR decision, an infrastructure change - the same question follows: what was done, under whose authority, and can it be proven later. And each profession needs a different cut of that same truth. A board, an auditor, an operations lead, a disputes team, and a regulator cannot usefully read the same raw stream. They need the same underlying record, framed for the desk they sit at.
AXES is the open standard for binding that evidence at the moment of execution, then rendering it as an Automation Execution Report each role could actually use. The situations it covers are not limited to financial services. The breadth of those use cases is wider than any single industry, and wider than any single product.
Two worked examples make that breadth concrete rather than asserted: a financial one, an autonomous accounts-payable run, and an industrial one, an autonomous production batch release. Both are built on the identical evidence skeleton. Finance came first because it sets one of the hardest evidentiary bars in practice, not because it is the boundary.
The same evidence skeleton, two industries. Open either example and you are looking at one record structure - authority, controls, commit, reconciliation, hash chain - doing its job in a different world. A payment instruction and a released machined part are the same shape of evidence. That is what makes AXES a standard rather than a format for one sector.
How it works.
- Capture once. Execution evidence is bound at the moment of action - authority, artefacts, and outcome - across whatever process the automation ran.
- Render for the role. Board, audit, forensic, regulator, conformance, and machine-readable views are cuts of the same record, not separate stories.
- Keep the report portable. The report is designed to become an organisation's own artefact. Branding belongs to the organisation that runs the process, not to the steward of the standard.
That is why a treasury lead, a fraud analyst, a plant operations manager, and a disputes investigator can open different views and still be looking at the same event.
Open standard. Conforming implementations.
AXES is developed in the open at github.com/magentixai/axes. Schema, documentation, and the Golden Trace evidence bundle are public. Requirements enter a public register that shows its working.
Magentix AI's execution-assurance product, ARBITR, is a conforming implementation of the kind of execution-evidence posture AXES describes. ARBITR is the first conforming implementation, which is how Magentix AI stress-tests the standard against real execution before asking anyone else to adopt it. AXES itself is not ARBITR, and it is not limited to Magentix AI products. Other implementations can conform. The standard is the shared language; the products are how organisations put that language to work.
Read an example. Shape the next draft.
The AXES Example publishes two worked evidence events, one financial and one industrial, each as several role-specific Automation Execution Reports. Choose the view that matches your world, read it as a report that crossed your desk, and answer what would need to change before you could rely on it. Your answers enter the public requirements register.