How Much Do AI Customer Service Agents Cost?

AI customer service agents typically cost $20 to $100 per user per month on a subscription, or roughly $0.05 to $0.50 per resolved conversation if you pay by usage. Outcome-based pricing, where you pay per resolved ticket, runs around $0.99 per resolution. A custom-built support agent is a different market entirely, commonly $20,000 or more to develop plus ongoing maintenance. For most teams, a ready-made subscription agent is the cheapest way to start, which is why buying beats building here.

That is the short answer. The rest of this guide explains what drives the number, which pricing model fits which team, and the trap that makes a cheap-looking agent expensive as you grow. If you want to see agents and their plans directly, you can browse vetted support agents or read the full AI agents for customer service breakdown.

The three ways support agents are priced

Subscription (per user or per month)

The most common model. You pay a flat monthly fee, often per human seat, and the agent handles unlimited or capped volume within that. Subscription is predictable, which finance teams like, and it is the easiest to budget. Ready-made support agents in this model generally land between $20 and $100 per user per month depending on features and volume. On Agentmarketplace, plans start at $29 a month.

Usage (per conversation or resolution)

You pay for what the agent actually handles, usually a few cents to half a dollar per conversation. Usage pricing is efficient when your volume is spiky or seasonal, because you are not paying for idle capacity. The risk is the opposite of subscription: a busy month is an expensive month, and a viral spike in tickets produces a bill to match.

Outcome (per resolved ticket)

The newest model, where you pay only when the agent fully resolves an issue without a human, often around $0.99 per resolution. It sounds fair, and for a steady, low-volume queue it can be. The catch is that it scales against you: the better the agent gets and the more it resolves, the larger your bill grows. Always model outcome pricing at several times your current ticket volume before you commit.

What actually drives the cost

Two agents at the same headline price can cost very different amounts in practice. The real drivers are:

• Ticket volume. The single biggest factor on usage and outcome plans. Estimate your monthly resolvable volume before comparing quotes.
• Channels. An agent covering chat, email, and voice costs more than a chat-only one.
• Integrations. Connecting to your help desk, order system, and payment tools is where the value is, and sometimes where the price is.
• How much it resolves versus deflects. A grounded agent that actually closes tickets is worth far more than one that only shows help articles.

That last point matters most. A cheap agent that deflects but rarely resolves just moves the ticket to a human later, so you pay twice. The full comparison of pricing models across every agent type is on our AI agent pricing page.

How to keep the bill from scaling against you

The teams that control support-agent cost do three things. They start with the highest-volume, lowest-risk ticket type so the agent resolves a large share of volume quickly. They keep a human on refunds and account changes so nothing expensive slips. And they route the tickets the agent should not touch to the right person instead of letting it guess, which is easier when you route each ticket to the right person automatically before it ever reaches a queue. Cleaner routing means the agent handles what it is good at and humans get only what needs judgment.

Is it cheaper to build or buy?

Buying is cheaper for almost every team. A ready-made agent costs tens of dollars a month and works this week. A custom build commonly runs five or six figures, takes months, and leaves you owning maintenance forever. Building only pays off when your support workflow is genuinely unique and you have the engineering team to keep it alive. We lay out both sides in buy vs build AI agents.

The bottom line

Budget $20 to $100 per user per month for a subscription support agent, or a few cents to a dollar per conversation on usage. Model any usage or outcome plan at several times your current volume, keep humans on the sensitive actions, and pick the agent that resolves rather than just deflects. When you are ready, see how support agents work and deploy one in a single click.