Understand your plan limits and entitlement meters
Minutes, messages, users, channels, automations and storage. What each one actually counts, and the arithmetic that surprises people.
Six things are limited and they are limited in two different ways, which is where the confusion starts. Users and channels are ceilings — a number you are at or below, right now. Minutes, messages and automation actions are monthly allowances that reset. Storage is a ceiling that only ever grows unless you delete something.
The arithmetic that catches people is automation actions, which are counted per step rather than per workflow run. A six-step workflow that runs a hundred times has used six hundred, not a hundred.
Before you start
- Admin access — Billing is admin-only.
- Five minutes a month. This is a habit rather than a task.
Reference
Find the meters in Billing
The subscription card shows users, channels, automations and telephony numbers, each reading "used N of M". Billing is admin-only, deliberately — it costs money and binds the company.
Checking these monthly is a two-minute habit that prevents a bad Tuesday.
A calling minute is AI agent time on a call
Inbound or outbound. Starter includes 600 a month, Growth 2,400, Scale 7,000 and Enterprise 20,000.
Time your team spends on calls is not metered. This is why putting an AI agent only on the after-hours branch is such an efficient first deployment — it uses minutes exactly where there was previously nobody at all.
Dialer minutes are the autodialler's carrier time, metered separately
When a campaign places calls, the carrier time it uses is counted as dialer minutes: Starter includes 1,200 a month, Growth 4,800, Scale 14,000 and Enterprise 40,000. This is a different meter from AI calling minutes — a campaign call draws on both, the dialer for the line and the AI agent for the conversation.
All call time is billed in 30-second pulses rather than rounded up to whole minutes, so a 40-second attempt costs two pulses, not two minutes. If a campaign is going to push you past either meter, extra minutes are available as an add-on rather than needing a new plan.
An AI message is one your bot sends on a chat channel
Starter 1,200 a month, Growth 4,800, Scale 14,000, Enterprise 40,000.
Messages your team types are not metered at all. So a busy shared inbox with no bot costs you nothing in messages, and pointing a high-volume channel at a bot is the thing that consumes them.
Users and channels are ceilings, not allowances
1 user and 3 channels on Starter; 3 and 4 on Growth; 10 and 7 on Scale; 20 and 12 on Enterprise.
All thirteen channel types are on every plan. Only how many you connect at once changes.
Automation actions are counted per step
5,000 a month on Starter, 15,000 on Growth, 50,000 on Scale, 100,000 on Enterprise.
Do the multiplication before you enable a workflow: steps × how often the trigger fires. A workflow on "every unassigned chat" costs your whole chat volume times its step count. Narrowing it with an "only if" is the cheapest optimisation available.
Storage is shared between four things
2GB on Starter, 14GB on Growth, 50GB on Scale, 120GB on Enterprise — covering knowledge base uploads, files on contacts, documents generated by workflows and call recordings.
Recordings are usually the largest consumer on a calling-heavy workspace. Knowledge bases are the best place to start clearing, because deleting superseded documents improves accuracy as well as space.
Change plan from the same pane
Plan changes, payment method, usage summary and cancellation all live on Billing.
Paying yearly costs ten months instead of twelve on every plan, which is often cheaper than moving up a tier for headroom you only need occasionally.
Put it to use
Work out your two largest monthly consumers and write them down. For most teams it is calling minutes and automation actions, and knowing which one you will hit first tells you which plan change to make when you grow.
The parts people get wrong
- Voice quality is not a plan difference — every plan gets the same ultra-realistic voices. What plans meter is AI voice minutes, which is time rather than quality. Voice is set per language on the bot.
- Every plan includes AI calling, the chatbot, the knowledge base, call flows, campaigns and all thirteen channel types. The tiers differ on volume — except white labelling (Scale and Enterprise) and advanced process settings (Growth upwards).
- A workflow that is switched off consumes no automation actions at all.
The ones people actually ask
What actually happens when I hit a limit?
It depends which one, and they behave differently. Out of AI messages, your bot stops sending while your team carries on unaffected. Out of AI voice minutes, your AI agent stops taking calls. Out of automation actions, workflows stop running. Out of seats or channels, you cannot add more until you free one. Calling balance is the exception and the serious one — it is money rather than an allowance, and it stops your phones outright.
Do the allowances reset?
AI voice minutes, AI messages and automation actions are monthly allowances that reset. Users and channels are ceilings rather than allowances — they do not reset because nothing accumulates. Calling balance neither resets nor accumulates: it runs down until you top it up.
Automation actions are climbing much faster than I expected. Why?
They are counted per step, not per workflow run. A six-step workflow running a hundred times a week costs six hundred actions, not a hundred. Narrowing a workflow with an only if is the cheapest fix available and usually improves its behaviour too; trimming a step nobody acts on is the second cheapest.
Is it cheaper to pay yearly?
Yes — the yearly price is exactly ten times the monthly one on every plan, so you pay for ten months instead of twelve. That is often cheaper than moving up a tier to get headroom you only need occasionally.