TIM rules on what your AI is about to do. Give it the action and the facts, and it answers allow, allow with conditions, hold, escalate or deny, and cites the rule it applied. Your models do the work. TIM decides when it's safe.
action release a $250,000 wire, one approver signed ruling HOLD rule Release of a customer wire · needs a second approver action charge an overdraft fee at an ATM, no opt-in ruling DENY rule Overdraft fee on a one-time debit or ATM transaction action agent signs and sends crypto to an unscreened wallet ruling DENY rule AI agent action on crypto-assets
Before an agent moves money, shares data or sends a notice, ask TIM. It applies the rule in force on the day and tells you which one it used, so a person can check it.
Nothing in the rule stops this action.
Go ahead once the rule's conditions are met, such as a notice or a report.
Wait: something must happen first, like a callback or a second approver.
A person decides. TIM hands off instead of guessing.
The rule does not allow it.
| A general model as judge | TIM | |
|---|---|---|
| Knows the rules | Only what you paste into every call | Built in, applied as written |
| Answers with | Free text you parse and hope | One of five rulings, and the rule it cites |
| When unsure | Sounds sure anyway | Escalates to a person |
| Rules change | Re-prompt and re-test | Applies the version in force on the day |
TIM is resting, try again shortly.
That doesn't match a rule this demo covers. Closest scenarios:
You've used your free decisions for today. Sign in with your HumaneCircle account to keep going.
Sign inThe playground is a demonstration on example bank rules. It isn't legal advice.
Beta pricing, matched to Jev: $0.042 per million input tokens; output is free. A typical decision is about 600 input tokens, so 1,000 decisions cost about $0.025.
Try it free on this page. API access opens to approved beta users; register interest below.
We're opening the API to a few beta users first. Tell us what you'd use TIM for.
We've got your details and will be in touch as beta places open.
No. TIM supports your compliance team; it doesn't replace it, and it isn't legal advice. Anything outside its rules goes to a person.
No. Keep the models you use for answers and agents. TIM rules on whether each action is allowed, and why.
It escalates to a person instead of guessing.
A selection of US banking, payments and AI rules, shown in plain words with each scenario. Each release comes with a coverage statement for your compliance team.
Not out of the box. We compile your policies in with you, then evaluate before it goes live.
For each decision: which rules were asked, the ruling, how long it took, and a one-way hash of your session or your account id, for abuse review. If you describe a situation, we keep up to 500 characters of it.
API access opens to approved beta users. Register interest above and we'll be in touch.
