Aircall publishes its UK AI Voice Agent rates, but a single headline price does not tell you what a month of calls will cost. The total depends on minutes, whether you buy a bundle, and whether the agent makes outbound calls. This guide turns the published rates into a practical budget and flags the charges that are easy to miss.

Reviewed 7 October 2026. Prices and product terms can change; check Aircall’s current quote and invoice before you commit.

Aircall AI Voice Agent prices in the UK

Aircall currently lists a free allowance of 50 minutes per month after 100 minutes on sign-up, pay-as-you-go at £0.37 per minute up to 2,500 minutes and £0.29 per minute from 2,500 minutes onward. Its published prepaid options include 500 minutes for £130 and 2,500 minutes for £550. The service has no contract, minimum usage or cancellation commitment on the UK product page. See the Aircall UK AI Voice Agent pricing page for the current offer.

Aircall bills participating AI calls to the precise second, expressed as a decimal minute. Calls shorter than 15 seconds are not billable. That differs from a rounded-up whole-minute model, so use the actual billed duration from your usage report when estimating. Aircall’s billing guide says its report in the dashboard is the source of truth for invoiced usage.

Example monthly costs, before phone-system charges

The examples below assume inbound AI participation, use the published tier rates, subtract the recurring 50-minute allowance, and exclude taxes, toll-free charges, any separately contracted Aircall seats and fees for outbound attempts. They are arithmetic illustrations, not quotes.

  • 500 AI minutes: 450 billable minutes × £0.37 = £166.50 at standard pay-as-you-go rates.
  • 1,000 AI minutes: 950 billable minutes × £0.37 = £351.50.
  • 3,000 AI minutes: 2,500 minutes at £0.37 plus 450 at £0.29 = £1,055.50.

The first month can differ because the sign-up offer includes 100 minutes. Verify the allowance shown on your account and how it is applied to the invoice.

When a prepaid bundle may make sense

A 500-minute bundle costs £130, or an effective £0.26 per included minute; the 2,500-minute bundle costs £550, or £0.22 per included minute. Those rates are lower than the standard pay-as-you-go tiers, but the bundle is prepaid. If usage is below the block you buy, the unused capacity can make the effective cost per used minute higher. Aircall says bundle credits offset usage charges, and are applied to the highest-priced tiers first. Confirm the specific terms in your order and billing view before treating the full allowance as rollover credit.

For example, 300 used minutes against a £130 / 500-minute bundle means £130 paid for 300 minutes of actual usage: an effective 43.3p per used minute. That is higher than the 26p headline rate because the full block was not consumed. A useful rule is to compare your recent monthly usage with the bundle size, rather than selecting a block from a busy week.

The outbound fee is separate from minutes

Every outbound AI Voice Agent call attempt currently adds a £0.011 origination fee, even if nobody answers or the call fails to connect. That fee sits on top of AI minutes and is not included in minute bundles, according to Aircall’s usage and charges documentation. A campaign with 1,000 attempts would therefore add £11 in origination fees before any billable conversation time. This is especially relevant when comparing inbound answering with outbound reminders or lead follow-up.

Aircall says local inbound telephony is included except for toll-free numbers; AI participation is charged separately from telephony. If your workflow uses toll-free numbers, international routing or outbound calling, ask for those line items separately rather than folding every cost into the AI per-minute rate.

A five-point Aircall cost check

  1. Estimate inbound and outbound agent minutes separately from human call minutes.
  2. Apply the free allowance once, then calculate each monthly pricing tier progressively.
  3. Add one origination fee for every outbound attempt, including unanswered attempts.
  4. Compare prepaid bundle size with normal usage over several months, not one campaign peak.
  5. Confirm toll-free, international, tax, number and phone-system seat charges on the final quote.

If you are comparing how AI phone systems fit into the rest of a customer workflow, see Sysevo’s AI voice agent overview and booking and calendar workflows.

Questions buyers ask

Does Aircall round every AI Voice Agent call up to a full minute?

No. Aircall says it bills participating calls to the exact decimal minute. Calls under 15 seconds are not billable. The usage report shows the billed duration used for invoicing.

Are the UK AI minutes included with an Aircall phone plan?

Aircall lists AI Voice Agent usage separately from telephony. Confirm the AI allowance and any phone-system or seat charges in your account quote.

Do prepaid bundles cover outbound call attempts?

No. Aircall documents an additional origination fee for each outbound attempt, including one that is unanswered or fails to connect. Minute bundles offset AI usage, not this per-attempt fee.

How should I choose between pay-as-you-go and a bundle?

Compare expected usage across several normal months with the bundle size and its prepaid cost. Include unused minutes and outbound attempts in the calculation, then verify the actual terms on your quote.

Aircall AI Voice Agent pricing for US buyers

This guide focuses on Aircall’s UK pricing because the headline rates and currency differ by market. US buyers should use the US rate card, not convert the UK price as a shortcut. Aircall’s US pages currently list 50 free AI Voice Agent minutes per month, a $0.49-per-minute tier through 2,500 minutes, and $0.39 per minute above that threshold. Its published US bundles include 500 minutes for $175 and 2,500 minutes for $725. Verify the live terms on the US page and your quote before budgeting.

The same usage-report rules apply as a practical budgeting concern: AI participation is billed separately from telephony, calls shorter than 15 seconds are not billable, and participating calls are counted to the exact second as decimal minutes. Do not assume those rules cover every phone charge. Toll-free traffic, international destinations, routing and outbound call attempts can have separate treatment. Ask Aircall to identify each line item for your specific US number and call flow.

For a basic US pay-as-you-go illustration, 1,000 total AI minutes with a 50-minute monthly allowance leaves 950 minutes in the paid tier: 950 × $0.49 = $465.50. At 3,000 total minutes, assuming the first 50 are free and the tier threshold applies progressively, the example would be 2,500 paid minutes at $0.49 plus 450 at $0.39, or $1,400. These calculations follow the published rate structure; confirm whether the threshold is applied exactly this way to your account and whether bundle credits alter the invoice.

Keep UK and US costs separate in a cross-border budget

If your company operates in both markets, make a separate row for each country. Record local currency, included minutes, rate tiers, bundle sizes, outbound attempt charges, toll-free treatment, number fees, taxes and the currency of the invoice. Do not add pounds and dollars or compare them without a dated exchange-rate assumption. A global team may use one account but still have different local rates and telephony rules.

Ask whether usage allowances are per account, per organization, per country or per agent. Confirm where a bundle applies and whether one market’s unused minutes can offset another market’s usage. Review whether phone numbers are local to the country where calls are answered and whether a caller crossing a border incurs a different charge. These questions are particularly important when one operations team handles both US and UK calls.

Build a more accurate UK monthly forecast

For the UK rates in the main guide, start with the last 30 to 90 days of calls and separate AI-active minutes from total phone minutes. Apply the 50-minute recurring allowance once, then apply the published usage tiers to the remaining minutes. If you have a prepaid bundle, model the bundle as prepaid cash and the invoice credit separately so that the same usage is not counted twice. Aircall’s billing help explains that bundle credits offset the highest-priced tiers first.

Count outbound attempts separately from conversations. The UK documentation currently lists a £0.011 origination fee per outbound attempt even when a call is unanswered or fails to connect. Ask whether your exact campaign and phone destination use that rate, and verify it in the billing report. For a reminder workflow, a high number of short unanswered attempts can make per-attempt charges more important than the connected-minute rate.

Use usage reports to reconcile your invoice

Assign one billing owner to download the usage report and compare its billed-duration field with the invoice. Filter for the AI voice-call entry type, group by tier and flag calls below the minimum threshold. Keep a copy for the billing period and note any contractual rate that differs from the public page. Aircall says its usage report is the source of truth for billed AI usage and that the dashboard retains a limited history, so export records on a regular schedule.

Compare the invoice with your operational dashboard, but do not assume the two reports answer the same question. Analytics can help monitor call outcomes; the billing report is what you need to reconcile charges. Investigate differences by call ID and date before changing the budget model.

For both markets, compare cost per verified outcome rather than only cost per AI minute. A completed booking, correctly routed lead or resolved question can be worth more operationally than a short but unsuccessful call. Recheck the UK product page, US pricing page and billing guide before procurement; country, contract and product terms can change.

How to choose a bundle without overbuying

Use at least three months of expected usage where available. For each month, apply the free allowance, tier prices, bundle credit and outbound attempt fees according to the current contract. Compare the prepaid cost with pay-as-you-go across low, expected and high usage. A bundle’s effective price per minute assumes the whole allowance is useful; if the business uses only part of it, calculate cost per minute actually consumed as well.

Include a month with a campaign or seasonal peak, but do not let a one-off event set the recurring bundle size without considering how often it happens. Ask whether a larger bundle can be added temporarily, whether it renews automatically and what happens to unused minutes. Keep the answer in writing because bundle treatment and contract-specific rates may differ from a public example.

Audit the outbound attempt count

For reminder or follow-up campaigns, estimate attempts separately from connected conversations. Export campaign logs and group them by answer, voicemail, busy signal, no answer and failed connection. Multiply the attempt count by the current origination fee, then add billable AI duration. This helps explain why an outbound workflow can cost more than its connected-minute total suggests.

Test retry rules and time windows carefully. A second attempt can improve reach but also add a fee and may create a poor caller experience if the contact has already responded through another channel. Establish a stop condition when a person replies, books or opts out, and reconcile campaign attempts with the vendor usage report.

Related Sysevo guides

Mapping a Full Cost Picture Before You Commit

The published per-minute rates and bundle prices are the starting point for an evaluation, not the finishing line. A business considering an AI voice agent for inbound enquiry handling or outbound follow-up will encounter several cost layers that interact with each other. Working through those layers in a structured way, before signing or scaling, reduces the risk of invoice surprises and makes like-for-like comparisons with alternative solutions more reliable.

The exercise that follows is framed as a practical evaluation method rather than a description of how any specific platform behaves. Treat each step as a question to put to the vendor and verify against your own account documentation.

A Worked Scenario: Mapping Costs for an Inbound Booking Line

Consider a fictional independent letting agency, Fernside Properties, that receives a steady volume of inbound calls asking about property viewings. Its team handles calls during office hours but misses calls in the evenings and on weekends. The business is evaluating whether an AI voice agent could handle out-of-hours enquiries, qualify callers and offer appointment slots.

At the evaluation stage, Fernside's operations manager pulls three months of phone records and filters for calls received outside staffed hours. She notes total call volume, average call duration and how many calls lasted fewer than fifteen seconds, which under the published billing rules would not be chargeable. She separates these figures from daytime calls handled by human staff, because only the AI-participated calls would attract AI usage charges.

She then applies the recurring free allowance to the monthly AI-minute estimate, uses the published tier rates progressively on the remainder, and records a figure for each of the three months rather than averaging them. One month contains a short property fair that generated a spike in after-hours enquiries. She notes that spike separately and asks herself whether it recurs seasonally, because including it in a bundle-size decision without that context could lead to overbuying for the other eleven months.

Fernside does not currently run outbound reminder calls, but the manager notes that the vendor documentation lists a per-attempt origination fee for outbound calls regardless of whether the call connects. She flags this as a line item to revisit if the agency later wants to send viewing reminders automatically, because the attempt count in a reminder campaign could be meaningfully higher than the count of connected conversations.

Distinguishing Customer Identity from Individual Requests

In a letting agency scenario, a single landlord or tenant may call several times across a billing period about the same property or booking. Each call is a separate billable event if the AI agent participates, but operationally those calls represent one customer relationship. When evaluating whether an AI agent delivers value, the relevant question is what happened across all contacts for that customer, not just the cost of any single call.

This distinction also matters for designing the agent's behaviour. If a caller has already booked a viewing and rings again with a follow-up question, the agent's response should ideally reflect the existing context. Evaluating how the system handles returning callers, and how that affects call duration and therefore billable minutes, is a practical acceptance criterion that belongs in any trial or pilot design.

Acceptance Exercise: Observable Evidence for a Pilot Period

Before committing to a recurring bundle or a full deployment, a structured pilot with observable acceptance criteria is more useful than a demo alone. The following criteria are illustrative and should be adapted to the specific use case.

  • Call routing accuracy: Does the agent correctly identify the caller's intent and route or respond accordingly, without requiring the caller to repeat information to a human agent?
  • Duration profile: Does the average AI-participated call duration during the pilot match the estimate used for budget modelling? A significant divergence in either direction requires explanation before scaling.
  • Short-call rate: What proportion of AI calls fall below the minimum billable threshold? A high rate may indicate callers abandoning quickly, which is an experience problem separate from cost.
  • Outbound attempt-to-connection ratio: If outbound calling is included, record attempts and connected calls separately. A low connection rate multiplied by the per-attempt fee affects the cost per useful outcome.
  • Billing report reconciliation: Export the usage report at the end of the pilot period and compare billed duration with the operational dashboard. Assign one team member to document any discrepancy and its resolution before the pilot concludes.

Handover Arrangements and Staff Responses

A voice agent handling inbound calls outside office hours will, in some scenarios, need to hand a call or a record to a human colleague the following morning. The handover design affects both caller experience and operational cost. Consider what information the agent captures, where it is stored, who reviews it, and how quickly a staff member can act on it.

In the Fernside scenario, a caller who requests a viewing outside office hours should ideally receive a confirmed time or a clear expectation of when they will hear back. If the agent collects a preferred slot but cannot confirm it in real time, the handover record needs enough detail for the lettings team to follow up without calling the customer back for information they already provided. Reviewing handover records from a pilot period is a practical way to assess whether the agent is reducing or redistributing staff effort.

Staff who receive handover records should have a defined response time and a clear escalation path for urgent or complex situations. An AI agent handling property enquiries may occasionally receive a call involving a maintenance emergency or a vulnerable caller. The operational policy for those cases needs to be documented separately from the AI agent configuration, and staff need to know what it is.

Questions That Should Change a Purchase Decision

Several questions, if answered differently than assumed, would materially alter the cost or suitability of an AI voice agent deployment. These are worth settling before signing rather than discovering on the first invoice.

  • Are unused bundle minutes carried forward, forfeited, or credited in some other way, and is that confirmed in the order documentation rather than only on the public product page?
  • Does the free monthly allowance apply from the first call of the month or after a threshold is reached, and how does the account dashboard display remaining allowance in real time?
  • If call volume drops significantly in a given month, is there a mechanism to adjust the bundle size, or is the prepaid amount fixed for a defined period?
  • What telephony charges apply to the specific phone numbers your business uses, including any numbers already in service that would be connected to the AI agent workflow?
  • If the business operates across more than one site or handles calls for more than one trading name, how does the account structure reflect that, and do usage allowances apply per account or per number?

Raising these questions with the vendor before procurement, and keeping written answers as part of the procurement record, gives a business a practical basis for reconciling future invoices and for renegotiating terms if usage patterns change substantially. Cost-per-outcome thinking, measuring what the agent achieves in completed bookings, resolved enquiries or qualified leads relative to total AI spend, tends to be more useful for ongoing evaluation than cost-per-minute alone, because it connects the billing report to the operational purpose the agent was deployed to serve.