Intelligence you can answer for: why every action has a named person

Every company in our category now sells intelligence. The large services firms put the word in their taglines, and the model makers measure it in benchmarks. That is fair: the models are genuinely capable, and getting more so every quarter. But intelligence is not what a chief financial officer, a chief risk officer or an auditor is short of. What they are short of is someone who can answer for what the intelligence did.

So when we chose the line that sits above all four elan1 platforms, we kept the word everyone uses and added the part nobody else can say. Intelligence you can answer for.

What "answer for" means in practice

Answering for an action is not a feeling of responsibility. It is three things you can point to after the fact:

1. 01A rule, written down before the action, that said whether this kind of action is allowed, needs a person or is refused.
2. 02A person, named before the action, who decided the cases the rule sent to them, and who could have said no.
3. 03A record, written at the moment of decision, that shows the rule, the person and the outcome, and that someone outside the company can check.

If any of the three is missing, nobody can answer for the action. They can only explain it, which is a weaker thing and a much longer meeting.

Why a person, and not a role

Most approval policies are written in roles: finance approves payments, credit approves overrides. That works when the people doing the work know who "finance" is this week. It fails when the work is done by an agent at two in the morning. A role cannot open a phone, read the reason an action was held and say yes. A person can.

That is why elan1 asks you to name approvers, with a deputy for each, and why an unanswered request expires after 24 hours and is refused rather than waved through. The name is what turns a policy into something an auditor can test. I have written separately about how to name an approver for every action an agent takes.

How the line reads across the four platforms

A masterbrand line has to fit every product, and the ones we have not built yet. We tested this one against each platform before we used it.

• elan1 Work: intelligence that drafts the change in your ERP or CRM, and a person you name who commits it.
• elan1 Decide: intelligence that makes the everyday decision, with a stated confidence and the reasoning to answer for it.
• elan1 Govern: the answer itself. Allow, ask a person, or refuse, with a record.
• elan1 Run: intelligence running in your country, on parts you can name, so you can answer for where it lives.

Our older line, "Every action, answerable", still has a job. It is the proof under the promise, and it is Govern's own line. On its own it described one product. The new line describes the company.

What we will not claim

A line about answerability has to be answerable too. So we do not describe elan1 as autonomous, and we do not say an agent "decides" anything that commits your business. We publish what is built and what is not: every product claim on this site carries its status, and the claims register lists each one as available, in pilot, designed or planned. Most of what this note describes is designed. When that changes, the register changes first.

We also have no customer results to show yet, and we will not borrow any. When we can show a governed action running in a real business, with the business's permission, we will. Until then, you can see a governed action at an illustrative distributor, from draft to record.