Resources · 2 min read
What a WhatsApp agent actually costs (and why “$0/month” is a lie)
The numbers change by country and scope. The price structure doesn't — and knowing it is what lets you compare quotes.
Every serious quote has two numbers
A one-time setup — wiring the agent to your catalog or system, defining what it does when it doesn't know, testing it against real scenarios — and a monthly fee that keeps the system alive: the model's API, hosting, monitoring, and adjustments when your business changes. If a proposal has only one of the two, half the story is missing.
Why “$0/month” is a red flag
Every conversation consumes model API. The server runs 24/7. Someone has to find out if the bot goes down on a Saturday — a dead agent doesn't complain, it just goes quiet. If there's no monthly fee, there are three options: it isn't AI (it's a button tree), the cost is hidden somewhere else, or nobody is going to maintain it. All three end the same way.
What actually moves the price
It isn't message volume: it's the work each answer does. An FAQ answerer and an agent that queries your live inventory, recommends, and escalates the delicate cases to a person aren't the same product at different prices — they're different products. The integration with your systems and the guardrails around the model are where the cost lives, and the value too.
The four questions that make quotes comparable
Ask for the setup vs. monthly breakdown. Ask who pays for the model's API and what happens to the cost if volume doubles. Ask for what “maintenance” includes, in writing. And ask how a change gets tested before your customers see it. Two quotes can only be compared once all four answers are on the table.
Rather go through it on your own case?
A 15-minute audit of your real operation. You leave with an honest read on what to automate first, and what to leave alone.
Book a 15-min audit