AI Credit & Spend Limits

How AI Credit Works

AI billing lets an installation charge each tenant for the AI it uses, from an allowance included with the tenant's plan and then from top-up credit, and stop a tenant's AI cleanly when both are used up. It is off unless you switch it on. It is intended for ActiveManage's hosted service and for installations that report their AI spend to ActiveManage; most single-organisation installations never need it.

Where to find it

Architect Panel → Configuration:

• AI Settings — the section AI credit and billing, where it is switched on and priced

Architect Panel → Subscriptions:

• AI Credit Ledger — every charge, top-up, refund and adjustment, per tenant
• AI Package Limits — per-package allowances and weekly caps, read-only here
• Tenant Plans — a single tenant's own AI allowance and weekly cap

Architect Panel → Activity:

• AI Spend Notices — the 80% and 100% e-mails, and who was told
• AI Credit Holds — credit reserved by AI calls still running

Admin Panel → AI:

• AI Credit — what a tenant's administrators see

What a call costs

Every AI call, whether through a language model, TypeSafe Jev, or image and video generation, is charged to the tenant it was made for at the provider's list price, converted with USD to GBP and multiplied by Markup. The charge is fixed when the call is logged, so changing the rate later never re-prices history. A model the price table does not know is charged at Rate for an unpriced model, so an unpriced model is never a cheap route. A call that never reached a provider costs nothing.

Where a tenant's allowance comes from

A tenant has an allowance for each billing period of its plan. The first of these that is set wins:

1. The tenant's own AI credit per period (pence) on Tenant Plans.
2. Its package's Included AI credit per period (pence) on AI Package Limits.
3. The plan's built-in allowance.

-1 means unlimited, and 0 means none (top-ups only). A weekly cap, Monday to Sunday UK time, is resolved the same way through AI weekly cap (pence) and Weekly AI cap (pence); 0 means no weekly cap, which is the default.

The order money is spent

• The plan allowance first, until it runs out or the weekly cap is reached.
• Then top-up credit, bought on the AI Credit screen, kept until used, and usable even once the weekly cap is hit.
• Anything neither covers is absorbed, recorded as Absorbed on the ledger and never charged back. This only happens when a call runs past its own estimate.

Who is charged

• A tenant's users: charged to their tenant.
• Architects: recorded, never charged and never refused.
• Tenant 0 (the sign-up wizard and platform background work): recorded at cost, never charged, and refused for the rest of the day once the day's total reaches Daily cap on tenant-0 AI (GBP).

With tenancy on, a tenant with no Tenant Plans row gets no AI at all, and a tenant whose subscription is not active is refused AI as well.

When credit runs out

The platform checks before it sends anything, so a refused call costs nothing. The person sees a plain sentence instead of an answer, for example that the plan's AI credit for the period is used up and when it resets, that this week's limit is reached, or that the request is larger than the credit left. Each refusal is still a row in the run log.

Calls already running hold their estimated cost, so two calls at once cannot spend the same credit. A hold is released when its call is settled, and stops counting after 15 minutes. While a tenant's remaining credit is below "Low" means below (GBP), only Calls at once while credit is low of its calls may run together.

Notices

At Warn at (80% by default) and at 100% of the allowance, and of a weekly cap, an e-mail goes once a period to up to ten of the tenant's administrators and to the ActiveManage notice address. Each is recorded on AI Spend Notices with who it went to and whether it was sent. If both are due together, only the 100% notice is sent.

Other effects while billing is on

• Log every call is forced on, because a call is charged from its log row.
• Send to AI's unmetered deliveries are withdrawn unless Unmetered Send to AI deliveries allows them.
• Amazon Bedrock no longer signs with the platform's own AWS keys unless you explicitly say so.
• The AI Builder can be opened to tenant administrators, if AI Builder for tenant administrators is on.

Worked example

A tenant's plan includes an allowance and no weekly cap. On the 20th its administrators receive the 80% notice. On the 26th a user's AI Builder request is refused with a sentence saying the period's credit is used up, when it resets, and that a top-up can be bought under AI Credit. The account owner buys one; the next request draws on the top-up, and the ledger shows the purchase and every charge after it.

Recommendations

• Leave billing off unless you genuinely charge tenants for AI.
• Let the package decide, and keep a tenant's own figures on Tenant Plans for real exceptions.
• Use a weekly cap where one heavy week could otherwise empty a month's allowance.
• Read AI Spend Notices to confirm warnings are actually being delivered.

Switching On AI Billing

Switching AI billing on is one setting, but on a multi-tenant installation it changes what every tenant can do from the moment it is saved. It is intended for ActiveManage's hosted service and for installations that report their AI spend to ActiveManage. This article is the checklist: what to prepare, which settings to fill in, and what to watch in the first days.

Where to find it

Architect Panel → Configuration:

• AI Settings — the section AI credit and billing

Architect Panel → Subscriptions:

• Tenant Plans — every tenant needs a row before billing goes on
• AI Package Limits — the allowances per package, read-only
• AI Credit Ledger — confirm charges are being written

Architect Panel → Activity:

• AI Spend Notices — confirm warnings reach people
• AI Credit Holds — credit reserved by calls in flight

Before you switch on

1. Check every live tenant has a plan. With tenancy on, open Tenant Plans and compare it with your tenants. A tenant without a row is refused AI as soon as billing is on.
2. Check the allowances. Each package's allowance and weekly cap comes from the plan catalogue unless AI Package Limits overrides it; that tile is read-only, and package limits are set by ActiveManage. For one tenant that should differ, set AI credit per period (pence) and AI weekly cap (pence) on its Tenant Plans row: blank keeps the package's figure, -1 is unlimited and 0 is none.
3. Choose who receives notices. Decide the operator address for the 80% and 100% notices, and which e-mail account sends them.
4. Check Bedrock credentials. If any role uses Amazon Bedrock with the platform's own AWS keys, give AI its own credentials first: with billing on, the platform keys are no longer used unless you say so explicitly.

The settings

• Charge AI to tenants: the switch.
• USD to GBP: the exchange rate applied to the providers' US dollar list prices. The shipped value is a placeholder: confirm it.
• Markup: multiplies the converted cost. Never below 1.
• Warn at: the fraction of the allowance that triggers the warning notice, between 0.5 and 0.99.
• ActiveManage notice address: the operator's address for every warning and 100% notice. Leave it empty and nobody on the operator's side is told.
• Send notices from e-mail account: 0 uses the password-reset account.
• Daily cap on tenant-0 AI (GBP): the limit on the sign-up wizard and platform work per UK day. 0 allows none, -1 is uncapped. The shipped value is a placeholder.
• Rate for an unpriced model: the charge for a model the price table lacks.
• Calls at once while credit is low and "Low" means below (GBP): stop a tenant with little credit left from running many calls together.
• AI Builder for tenant administrators: lets tenant administrators use the AI Builder on their own tenant.
• Unmetered Send to AI deliveries: Default withdraws them while billing is on. Leave the remaining settings at their defaults unless advised.

Switching on

1. Fill in the settings above on AI Settings and press Save on the billing section, with Charge AI to tenants on.
2. Make a small AI request as an ordinary user of a test tenant.
3. Open AI Credit Ledger and find the row: Kind usage, the tenant, Cost (USD), Amount (GBP) and From plan allowance.
4. Sign in as that tenant's administrator and open AI Credit in the Admin Panel to see the allowance bar move.
5. Make sure tenant administrators can see the AI Credit tile: it is granted automatically to self-serve tenants' administrator groups, and has to be granted to other groups.

What to watch afterwards

• AI Spend Notices: each row shows the scope, the threshold, the amounts, Sent to and Status (pending, sent, partial, failed, skipped or superseded). skipped means there was nobody to tell: check the notice address and that the tenant's administrators have e-mail addresses.
• AI Credit Holds: rows appear and disappear as calls start and finish. A hold stops counting after 15 minutes even if its call never settles.
• AI Credit Ledger: read-only, and never pruned. Rows of Kind purchase, refund, adjust and netoff record top-ups, refunds or disputes, staff corrections and a debt being repaid from the allowance.

What goes wrong, and how to tell

• A tenant reports "AI isn't available because this account's subscription isn't active": it has no plan row, or its subscription is not active.
• Every tenant is refused with "its AI credit couldn't be checked": the credit tables are not installed on this database. Ask your hosting administrator to apply the platform update.
• The sign-up wizard says "AI suggestions are busy right now": the tenant-0 daily cap has been reached.
• Bedrock roles start failing with not_configured: they relied on the platform's AWS keys. Give Bedrock its own credentials.

Worked example

An installation with twelve tenants reports its AI spend to ActiveManage and is to start charging tenants for AI. The architect checks Tenant Plans has a row for each tenant, adds two missing ones, and gives one tenant a larger allowance on its own row. On AI Settings they confirm the exchange rate, set the support team's address as the notice address, and switch billing on. A test request made as an ordinary user appears on the ledger straight away, and the first 80% notice a fortnight later shows on AI Spend Notices as sent to the tenant's administrators and to the support team.

Recommendations

• Do the Tenant Plans check first; it is the step that cuts tenants off if missed.
• Replace both placeholder figures before going live.
• Always set a notice address, or nobody on your side hears about a tenant running out.
• Test as an ordinary user: architects are never charged.

The AI Credit Screen

Where AI billing is on, a tenant's administrators can see their AI credit on one screen: how much of this period's allowance is used, the weekly limit if there is one, the top-up balance, which features spent it, and why AI is refused if it is. The person who manages the subscription can buy top-ups there.

Where to find it

Admin Panel → AI:

• AI Credit — the tenant's allowance, usage by feature, top-up balance and top-ups
• AI Builder — shows the credit left under its heading, with a link here

Admin Panel → Account:

• Billing & Plan — the plan and subscription itself

Who can open it

Administrators of the tenant they are signed in to, and architects. Anyone else is told "Only this app's administrators can see its AI credit." Where billing is off the screen says "AI use is not metered on this installation, so there is no AI credit to show."

Reading the screen

• The heading line: the plan, its status, and the dates of the current billing period.
• A yellow panel, when AI is currently refused, with the same sentence people see when an AI request is turned down.
• Included with your plan: a bar and a line giving the amount used, the allowance, what is left and the date it resets. The bar turns amber at the warning point and red when it is used up. An unlimited plan shows only what has been used.
• This week, when a weekly limit applies: used against the week's limit, resetting on Monday.
• Top-up credit: the balance, kept until it is used and spent once the plan's credit runs out.
• This period by feature: requests and amount charged for each feature, such as AI Builder, Query assistant, Code assistant, Document reading, Bank feed coding, Agent tasks, Decisions (Jev), Images and video.
• Top-ups, refunds and adjustments: the most recent movements other than usage, with what each was and any note.

Buying a top-up

1. On a self-serve account, sign in as the person who manages the subscription. Only they see the buy buttons; other administrators see "Ask the person who manages this account's subscription to buy a top-up."
2. Open AI Credit and press one of the Buy £… top-up buttons.
3. Pay on the payment page.
4. Back on the screen, "Payment received" shows until the payment is confirmed, then "Your top-up has arrived." The balance updates, and the purchase appears under Top-ups, refunds and adjustments.

If the payment page is abandoned, the screen says the top-up was cancelled and nothing has been charged.

Owed credit

If a top-up is refunded or disputed after its credit was used, the balance goes below zero. A yellow panel above the buy buttons explains the amount owed: it is taken off the plan's allowance when there is allowance to spare, and a new top-up pays it off first. On the movements list it appears as "Owed credit taken off the plan allowance".

For architects

An architect sees the screen read-only, with no buy buttons, and an ActiveManage staff card beside it. The card shows the value of staff AI use this period, which is recorded and never charged. Under Adjust the top-up balance, enter a positive or negative amount and a note, then press Record adjustment to correct a tenant's balance by up to £1,000 either way. Every adjustment is a row on the AI Credit Ledger.

What goes wrong, and how to tell

• "AI credit belongs to an app": you opened the screen outside any tenant. Open it from inside the app; an architect also sees the day's sign-up and platform AI against its cap.
• No buy buttons for an administrator: only the person who manages a self-serve account's subscription can buy.
• "Payment received" never turns into "arrived": the payment provider has not confirmed the payment yet. Reload later; the screen never adds credit on the strength of the return page alone.
• AI is refused although the bar shows credit left: the weekly limit has been reached, or the remaining credit is reserved by requests still running. The sentence shown with the refusal says which.

Worked example

A tenant's administrator is told by a colleague that the AI Builder's Ask button is greyed out. They open AI Credit and see the weekly bar full and a yellow panel saying this week's AI limit is reached, resetting on Monday, with top-up credit usable now. The account owner buys a top-up; shortly afterwards "Your top-up has arrived" shows, and the AI Builder works again on the top-up balance.

Recommendations

• Check This period by feature before buying more: it shows what is actually spending.
• Make sure the subscription owner knows they are the one who can buy.
• Read the yellow panel before reporting a fault: a refusal is usually the limit working.
• Architects: always add a note to an adjustment.

Reporting AI Spend to ActiveManage

An installation can send ActiveManage an hourly summary of its AI use, and ActiveManage can set a monthly and a weekly AI spend cap for that installation, which the installation then enforces itself. This is how AI spend is watched across separately hosted client installations as well as ActiveManage's own service.

Where to find it

Architect Panel → Activity:

• AI Usage — the Reporting to ActiveManage card, shown once the installation is enrolled
• AI Spend Notices — the installation-cap warnings

Architect Panel → Automation:

• Tasks — AI Spend Monitor - Report to ActiveManage, the hourly report

The AI Spend Monitor tile itself is ActiveManage's own console and appears only on ActiveManage's master installation.

What is sent, and what is not

• Sent: for each day, tenant, feature and model, the number of calls and errors, the tokens, the provider cost estimate and any charges; each tenant's code, name, plan and credit state; and the platform version.
• Never sent: prompts, answers, request or response content, who made a call, or any key.

Tenant names do leave the installation, so they are worth a line in your data processing agreement with ActiveManage.

Enrolling an installation

1. ActiveManage registers the installation and issues a reporting key, shown once as two lines of configuration.
2. Your hosting administrator adds those two lines to the installation's server configuration file. They are deliberately not a Site Settings value, so the key cannot be read or changed from a screen.
3. In Tasks, switch on AI Spend Monitor - Report to ActiveManage. It ships disabled and runs hourly.
4. Within the hour, open AI Usage and check the Reporting to ActiveManage card shows a Last accepted report and no Last error.
5. Make sure ActiveManage notice address and Send notices from e-mail account are set on AI Settings, so cap warnings go somewhere.

The first report fills in roughly the previous five weeks. With AI billing off the installation still reports usage, just with no charges.

The installation cap

The card shows Monthly AI cap and Weekly AI cap, each as used against the cap with a bar, and the caps' version, when they were received and their state. The month is the UK calendar month and the week Monday to Sunday, UK time.

• Below 80%: AI works normally.
• At 80%: one notice per month or week to the installation's platform administrators and the notice address, recorded on AI Spend Notices.
• At 100%: every AI call is refused before it is sent, with a sentence saying the installation has reached the AI spend limit set by ActiveManage.

The cap counts all AI use, architects and sign-up work included, and is enforced even when AI billing is off, measured then from the run log's cost estimates at the billing rate and markup. There is no top-up for an installation cap: ActiveManage raises it. Caps stay in force while ActiveManage cannot be reached, and stop applying once the installation is no longer enrolled.

What goes wrong, and how to tell

• The card says "not enrolled - the caps below are not in force": the configuration lines are missing or blank. Caps received earlier are still displayed, but they no longer apply.
• "the reporting key in config/config.php is not valid": the key line was copied wrongly. Ask ActiveManage to rotate it and paste the new line.
• Last error mentions a signature mismatch: the key does not match ActiveManage's record, usually after a rotation. Paste the new key.
• Last error mentions clock skew: the server's clock is more than five minutes out. Your hosting administrator should fix its time synchronisation.
• Reports refused as switched off: ActiveManage has paused reports from this installation; it retries after a day.
• Never reporting: the task is not switched on, the task engine is not running, or the two lines are missing.

After a failure the installation backs off and tries again later; Next attempt on the card says when, and how many failures there have been in a row.

Worked example

A client installation is enrolled and given a monthly cap. On the 22nd its architects receive an 80% notice. They open AI Usage, group usage by feature, and find a custom app's background job running far more often than intended. The developer fixes the schedule, spend levels off, and the cap is never reached. Had it been, AI would have stopped with a plain explanation until ActiveManage raised the cap or the month ended.

Recommendations

• Enrol before switching AI on widely, so spend is visible from the first day.
• Set the notice address, or the 80% warning reaches nobody on your side.
• Check the card after any server move; clock and key problems show there first.
• Agree the cap with ActiveManage before a known busy period, rather than after it is hit.