Check the numbers. Apply the rules.
Rules and policy
Calculate totals, test credit limits and check approval requirements against the relevant records. Cobalt shows the rule and figures behind the result.
Why is this order on hold?
The account balance, open commitments and proposed order are checked against the configured credit limit. The result identifies the rule that needs attention.
Why does the new order need review?
The credit check uses the configured limit and current exposure.
Rules and policyIllustrative records| Input | Amount |
|---|---|
| Credit limit | £20,000 |
| Existing exposure | £17,500 |
| Proposed order | £4,000 |
| Resulting exposure | £21,500 |
The order would exceed the limit by £1,500.
Show the relevant rule and route the exception through the configured approval process.
- It calculates, it doesn’t estimate. Margin, tax, remaining credit and how long customers take to pay are all sums over records you can open.
- It gives reasons, not scores. When a check fails, you see the rule that failed it, so “why?” always has a straight answer.
- It answers simple what-ifs. Change a quantity or a date and the same rules run again on the new figures. The same change always gives the same answer.
- It checks everything else first. Every suggestion passes through these rules before it is ranked, so you are never offered something that breaks them.
- It uses no AI credits. Account warnings, explanations and the month-end checklist keep working whatever your credit balance is.
A warning, a yes or no, or a set of limits.
Warning
Something that is true about a record right now, the rule that spotted it, and the figures it looked at.
Yes or no
Allowed or refused, the rule that decided it, and what would need to change for a different answer.
Limits
The hard limits everything else has to work within: capacity, budget, approval limits and closed months.
It only uses rules you already have.
It can’t make up a rule. It applies the ones already set up for your business, so when it refuses something you can go and read the rule behind it. If Cobalt ERP spots a pattern that suggests a new rule, a person has to approve it. It never becomes a rule on its own.
In every part of the system.
It is the main tool in all fourteen business areas, and the only part of Cobalt ERP Intelligence that is. Warnings on orders and invoices, the month-end checklist, explanations of where a price came from, clashes in who can approve what, budget variances and reorder alerts all come from here.
- “Why is this customer on hold?” — the credit rule that put them there.
- “Which rule set this price?” — the price list, the level it came from, and the line it applied to.
- “Can I close the month?” — the unposted documents, unmatched bank lines and outstanding approvals in the way.
- “Why can’t I cancel this order?” — the invoice already raised, the dispatch already booked and the serial-numbered stock already issued.
Frequently asked questions.
Is this really AI?
It uses no AI model, and it does most of the useful work. Every figure on screen is labelled with how it was worked out, so you always know whether a number was calculated by a rule or estimated by AI.
Does it cost credits?
No. Nothing here uses a paid AI model, so it keeps working when your credit balance is zero.
Can it be wrong?
If it’s given the wrong rule, it will apply that rule faithfully. What it won’t do is give two different answers to the same question on the same figures, which is what matters most in your accounts.
Why does everything else go through it first?
Because showing you an option that breaks your rules is worse than not showing it at all. The rules are always checked before options are ranked, and the system won’t accept a setup that does it the other way round.
Try it on your own records.
Book a demo and ask it a real question about your business. Every answer shows how it was worked out and which records it used.
Book a demo