All articles

BYO API Keys: How to Cut Your AI Calling Costs by 60%

· 5 min read

AI-powered calling features are genuinely useful. Transcription, translation, voice cloning, AI agents that can handle calls on your behalf — these are real tools that save real time. But they also consume resources, and those resources cost money.

ShoutDial tracks that consumption through something called AI minutes. If you have been watching your AI minute balance drop faster than expected, this post explains why — and how BYO API keys can cut that consumption dramatically.

What AI minutes are and why they have multipliers

Every ShoutDial plan includes a pool of AI minutes. Think of them as a shared budget for all the AI features you use during calls. A basic transcription job uses AI minutes. A translated call uses more. An AI agent handling an entire conversation uses the most.

The reason is straightforward: different AI features have different computational costs. A simple transcription runs through one model and produces text. An AI agent, on the other hand, is running a large language model in real time, generating responses, and potentially coordinating with text-to-speech — all while the caller waits on the line. The infrastructure cost per minute is not the same.

To keep pricing simple, ShoutDial uses a multiplier system instead of charging separately for each feature. Every minute of call time is multiplied by a factor based on which AI features are active. The highest multiplier wins — they do not stack.

The multiplier table

Here is the full breakdown of AI minute multipliers:

So if you run a 10-minute call with an AI agent active, that consumes 50 AI minutes from your pool (10 x 5.0). The same 10-minute call with just transcription would only consume 15 AI minutes (10 x 1.5).

Again, multipliers do not stack. If a call uses both transcription and an AI agent, you pay the 5.0x rate — not 6.5x. The highest active multiplier is the one that applies.

What BYO API keys means

BYO stands for Bring Your Own. ShoutDial lets you plug in your own API keys from two providers:

When you supply your own key, the API calls for that provider are billed directly to your account with that provider. ShoutDial still handles the orchestration, call routing, and integration — but because the heavy compute is running on your own API account, the AI minute multiplier drops significantly.

This is not a workaround or a hack. It is a built-in feature designed for users who already have accounts with these providers or who want more control over their AI costs.

How much you actually save

The numbers are concrete. Here is what happens to your multipliers when you bring your own keys:

Let us walk through an example. Say you run 100 minutes of AI agent calls in a month.

Without BYO keys: 100 minutes x 5.0 = 500 AI minutes consumed. Your Pro plan includes 120 AI minutes, so you need 380 additional minutes. At the best add-on rate (+60 AI Min for $30), that is roughly 7 packs = $210 in add-ons alone.

With your own Anthropic key: 100 minutes x 2.0 = 200 AI minutes consumed. You need 80 additional minutes beyond your Pro plan allocation. That is about 2 add-on packs = $60. Plus whatever Anthropic charges you directly for the API usage, which for Claude is typically a fraction of the managed rate.

The savings scale linearly. The more AI agent minutes you use, the bigger the gap.

The same math applies to ElevenLabs voice cloning. A 50-minute month of voice clone calls goes from 200 AI minutes (50 x 4.0) down to 75 AI minutes (50 x 1.5) — well within your included allocation on a Pro plan, with no add-ons needed at all.

Who benefits most

BYO keys make the most sense for two types of users:

Heavy AI agent users. If your business relies on Claude-powered agents to handle inbound calls — appointment scheduling, customer support triage, after-hours reception — those 5.0x multiplier minutes add up fast. Dropping to 2.0x is the single biggest cost lever available to you.

Voice cloning users. If you use ElevenLabs voice clones for branded IVR, outbound messaging, or any TTS that needs to sound like a specific person, the 4.0x multiplier makes it expensive at scale. With your own ElevenLabs key, it drops to the same rate as basic transcription.

There is also a third category: teams who already have Anthropic or ElevenLabs accounts for other projects. If you are paying for API access anyway — for a chatbot, a content tool, an internal app — plugging those same keys into ShoutDial costs you nothing extra on the key side while cutting your AI minute consumption substantially.

How to set it up

The setup takes about two minutes. In the ShoutDial portal, navigate to your account settings and find the API Keys section. Paste in your Anthropic API key, your ElevenLabs API key, or both. Save. That is it.

ShoutDial validates the key on save, so you will know immediately if something is wrong. Once the key is active, all subsequent calls using that provider will automatically use your key and apply the reduced multiplier. There is no need to change anything about how you make or receive calls.

You can remove or rotate your keys at any time. If you remove a key, calls revert to using ShoutDial-managed API access at the standard multiplier rate.

Which plans support BYO keys

BYO API keys are available on two plans:

If you need additional AI minutes beyond what BYO savings provide, add-on packs are available: +10 AI minutes for $8, +25 for $15, or +60 for $30.

BYO API keys are one of those features that quietly saves significant money for the people who need it. If you are running AI agents or voice clones at any meaningful volume, it is worth the five minutes to set up. Reach out if you have questions about how it would work for your specific setup, or log into the ShoutDial portal and add your keys now.

Ready to try ShoutDial?

AI-powered voice platform — your AI agent answers calls 24/7 in 30+ languages, in your voice. Starting at $49/month.

Start Free Trial