For the people accountable when AI publishes
Rules on paper are hope.Rules in Cortex are policy.
Generic AI writes with total confidence and zero memory of what your company has approved. Cortex is the governed memory inside arbitr — the reason every AI and human change gets checked against your rules, your claims, your markets, before it reaches production.
If you own the risk when something wrong goes live — the rewritten claim, the broken price, the off-brand launch — this page is for you.
Your rules already exist. They just can't act.
Brand guidelines in a PDF. Approved claims in a legal folder. Pricing rules in a spreadsheet. Market requirements in one reviewer's head. Every organisation has the knowledge — as documents.
And a document can't check anything.
brand-guidelines-v7-FINAL.pdf
last opened 214 days ago
Unenforceableapproved-claims-2024.docx
superseded, still circulating
Unenforceablemarket-rules-DE.xlsx
owner left in March
UnenforceableThis gap was survivable when content moved at human speed. AI publishing ended that. One prompt can now change a claim, a price, and a locale link across a thousand pages — faster than anyone re-reads the PDF. The rules didn't fail. They were never able to act.
Ten domains. Ten disasters that don't happen.
This is what a Cortex looks like from the inside — here, an illustrative workspace for a global brand publishing into the Japanese market. Every node is a body of approved knowledge. Every line is a relationship arbitr checks across.
Drag the nodes · click one to see what it prevents
Financial (J-GAAP)
412 entriesApproved J-GAAP reporting terminology and the English renderings your finance team signed off — operating income is not "profit," and 経常利益 has exactly one approved translation.
"operating income" → "profit"
Earnings summary · JA → EN translation
HeldRouted to Finance
Eliminates: Line-by-line re-review of every translated filing.
You buy arbitr. Cortex is why it works.
One clarification, so there's no mystery: arbitr is the platform — it connects to your CMS, email, and translation systems and checks every change before it publishes. Cortex is its memory — the governed store of everything your organisation has approved. arbitr decides; Cortex is what it decides with.
A change arrives.
Human or AI, from any connected system — arbitr sees it before it ships.
Cortex is the measure.
The change is compared against your approved phrases, locked fields, brand terms, and market rules — not a generic model's opinion.
The decision comes back — and stays.
Safe changes publish. Risky ones are corrected, held, or routed with the reason attached. Approved corrections flow back into Cortex, so the next thousand checks are sharper than the last.
Customer-controlled
Cortex is your asset. Your rules, your approvals, your precedents.
Human-approved
Nothing enters Cortex without a reviewer's decision.
Full audit trail
What changed, why it was held, who approved it — every decision, on the record.
A generic model can't do this. Structurally.
Not because it isn't smart. Because it's missing three things no prompt can add:
It has no memory of your approvals.
A general model knows the internet. It does not know that your legal team locked "clinically proven" in March, that /de/pricing must keep its locale path, or that your trial is 14 days — not the 30 it just confidently wrote. Paste your guidelines into a prompt and they last exactly one conversation.
It sits in the wrong place.
A chat window is upstream of the risk. Cortex sits inside arbitr at the publish boundary — the last moment before a change reaches your CMS, your email platform, your checkout. Advice before the fact is a suggestion. A check at the gate is a control.
It leaves no record.
When a regulator, an auditor, or your own board asks "who approved this?" — a chat log is not an answer. Every arbitr decision carries the rule that triggered it, the reviewer who resolved it, and the evidence behind it.
Generic AI generates. Risk owners need evidence.
“arbitr gave our customers better outputs with greater precision. To say I am impressed is an understatement.”
Hope is not a control.
Your rules are already written. Cortex is where they become enforceable — and arbitr is how they act, on every change, before it ships.