Glossary · Conversation
How does a prompt-based agent run a call?
Prompt-based agent
A prompt-based agent runs the whole call from one written brief: who it is, what it knows, and what it should get done. The language model decides each reply from that brief and the conversation so far.
On a call
A clinic writes one prompt: you're Riya from the front desk, confirm {{patient_name}}'s appointment for {{date}}, offer to move it if needed, and never give medical advice. Every reminder call runs from that page.
AI agents from ₹1 a minute.¹
Explained
Here is how it works, in plain words.
Most voice agents start life as a prompt. You write what the agent is, what it may say, and what a good call looks like. The model reads that before every reply and improvises the rest.
That’s a real strength. Callers go off script all the time: they ask about a refund in the middle of a delivery call, or switch from Hindi to English halfway through. A prompt-based agent can follow them, because nothing in it says the call must go step by step.
It works best when the call does one job, like confirming an order or reminding someone about a payment. A page or two of instructions covers that, and the model handles the small talk around it.
Writing for voice is different from writing for chat. The caller can’t scroll back, so the agent should say one thing at a time and ask one question per turn. Amounts and dates belong in the prompt as words, the way they should sound. Long lists of rules slow the model down, and a short prompt with one worked example tends to sound more natural.
The weak spot shows up as calls get complicated. When there are many branches, such as wrong person, reschedule, refusal or a question that must go to a human, a single prompt starts to lose its place. You also can’t easily see which step a call reached. That’s the point where teams move to a node-based agent.
Whichever you choose, test with real voices. A prompt that reads well on screen can still sound stiff on a call.
In Frequency
This is what Frequency does about it.
Every plan includes prompt-based and node-based agents, and the price is the same either way. A prompt can run up to 12,000 characters, with {{variables}} filled in for each call from your CSV or API request. Our prompting guide shows the shape we use: a role line, the facts, a short call flow and an example exchange.