Bill voice AI bots per minute and see the margin on every call
Your customers pay per minute, rounded up to 15, 30, or 60 seconds. PortaAIM tracks what every speech, language, and voice model behind those minutes costs you.
See it on your pricingOne bot call, up to 10 meters
The customer sees one call. Your vendors bill you for each part of it in their own units.
A 75-second bot call runs speech-to-text (STT) for all 75 seconds. The language model (LLM) bills tokens each time the caller finishes speaking. Text-to-speech (TTS) bills characters each time the bot answers. Around them sit telephony and smaller models like noise filtering, voice detection, end-of-turn detection, language detection, and interruption detection, plus knowledge-base lookups.
While some of these cost nothing today, they may cost something next quarter. Some use tokens even with no call, such as writing a prompt or summarizing a call.
Where per-minute billing loses track of cost
A flat price per minute is easy to sell. The cost behind it changes with every model your customers pick.
Today
Billing by the minute, costs by the token
- Customers pay per minute, vendors bill per token or second
- Margin is a guess based on average model costs
- A customer picks a pricier LLM and the margin shrinks
- A campaign keeps calling after the balance hits zero
- Finance builds invoices by hand from payment receipts
With PortaAIM
To price each part of the call, PortaAIM:
- PortaAIM prices each vendor unit and charges the minutes you report
- PortaAIM shows revenue, cost, and margin for each call
- PortaAIM prices each model and unlocks extra models per account
- PortaAIM checks the balance before each call starts
- PortaAIM issues invoices on demand or at month end, with tax
One 75-second call, priced and costed
Telephony
75 seconds
$0.010
STT
75 seconds of audio
$0.008
LLM
2,400 tokens in, 300 tokens out
$0.012
TTS
900 characters
$0.025
Your cost for the call
$0.055
Customer price
2 minutes (75 seconds rounded up to 60-second increments) at $0.10
$0.200
Margin on this call
$0.145
Example with round numbers. Your vendor prices and rounding will differ.
How PortaAIM bills a bot call
Your orchestrator talks to PortaAIM twice per call: before it starts and after it ends.
1
Load vendor prices
Enter each vendor's price in the unit it bills you: per hour, per minute, per 1,000 characters, or per million tokens.
2
Set customer plans
Set the price per minute for each plan, with a separate price for each model a customer can choose.
3
Check before the call
PortaAIM checks the balance and reserves the amount your platform asks for before the call starts.
4
Report after the call
Your orchestrator sends the billable minutes and each model's usage, one event per unit, under one call ID.
5
Read the result
PortaAIM books the charge and records revenue, cost, and margin under the call ID.
Your platform decides how to round each call, for example, up to the next 15, 30, or 60 seconds. PortaAIM charges the minutes it receives and tracks the real underlying cost.
Map your call flow to PortaAIMWhat you decide
PortaAIM holds the prices, balances, and invoices. Your platform keeps its users, bots, and logs.
Offer plans and a free tier
Run paid and free plans, each with its own PortaAIM per-minute prices.
Unlock models per account
Give one account a premium model without creating a new plan in PortaAIM.
Sell prepaid or postpaid
Take payment before usage or invoice after it, from the same PortaAIM setup.
Keep your payment gateway
Your gateway collects the money. PortaAIM records the payment and updates the balance.
Keep your observability tools
Transcripts and latency stay in your tools. Usage reaches PortaAIM through its API.
Tested in a pilot with Acefone
Acefone, an Indian cloud communications provider, runs a voice AI platform where businesses build bots on STT, LLM, and TTS models and pay a flat rate per pulse of call time.
The PortaAIM pilot aimed to let customers switch models from call to call while keeping pulse pricing, with the vendor cost of every STT, LLM, and TTS request tracked underneath.
“The PortaAIM team's support has been excellent. They took the time to understand our Voice AI business, from per-service billing for STT, LLM, TTS and telephony to in-country hosting in India. The platform clearly fits the kind of billing complexity we expect to face down the line.”
What it takes to start
A charging integration starts with two API calls from your orchestrator.
Two API calls to charge usage: check the balance before the call, report usage after it
Calls to create and read customers and to read plans, for a working setup
A sandbox hosted by PortaOne, to try PortaAIM on your own prices
Production in your country, your own environment, or as an on-premise (self-hosted) installation
FAQ
How do we meter and bill what each customer's voice AI bots across different STT, LLM, and TTS providers?
Your orchestrator reports each model's usage to PortaAIM under the call ID, one event per unit (seconds, tokens, or characters). PortaAIM prices each event at the vendor's rate and charges the customer for the minutes on their plan.
Can AI bot and human agent minutes from one call go on one invoice?
Yes. When your platform reports both to PortaAIM as separate services, each gets its own line on the same PortaAIM invoice.
Can customers pay per minute while we pay vendors per token?
Yes. In PortaAIM, the minute is the item the customer pays for, and the tokens, seconds, and characters behind it are tracked as your cost.
What happens when a customer's balance runs out during a campaign?
PortaAIM checks the balance before each call starts. When the balance or credit limit runs out, PortaAIM stops authorizing new calls until the customer tops up.
Can billing run in India?
Yes. PortaAIM can run in-country, in your own environment, or as an on-premise (self-hosted) installation, and PortaOne has already set up a PortaAIM installation in India.
Do we have to replace our observability stack?
No. Transcripts, latency, and event logs stay in your current tools. PortaAIM receives only the usage it needs to charge, through its API.
Reselling ready-made AI voice agents to business customers on PortaSwitch? See Voice AI Agents (Apifonica). This page is for platforms that build their own voice AI bots and bill them.
PortaAIM product page · Voice AI Agents (Apifonica)
