What AI customer support actually does
It answers the same forty questions a support team fields every day, using your help centre, policies and product docs as its source. It cites the article it used, hands anything about refunds, accounts or money to a human, and gets better as the knowledge base grows. It does not replace agents; it removes the tier-1 volume that buries them.
Why per-resolution pricing is winning
Support work is measurable, so it can be priced by result. The market rate is about ninety-nine cents per resolved ticket, often with a base platform fee on top. Outcome pricing removes direct price comparison and aligns you with the buyer: if the AI doesn't resolve, the client doesn't pay. That is why 43% of software companies now run a hybrid model, heading past 60%.
What it saves a business
A loaded support agent costs $35,000–$80,000 a year. Deflecting 70–85% of tier-1 tickets removes a large share of that volume from the queue, and the base-plus-usage bill scales only as value is delivered. For a busy desk, the monthly invoice is a fraction of the labour it replaces.
How to deploy it without risk
Start with a shadow period: let the agent draft while a human reviews every response, then step back as quality holds. Put a hard cap on spend, escalate anything touching money or accounts, and log every action and its source. A weekly scorecard of tasks, correct answers, escalations and failures is how you decide when to give it more.
What to measure
Deflection rate, resolution quality, and escalation rate — not the number of conversations. If deflection is well below 70%, the problem is almost always the knowledge base, not the model. Fix the content, retest, and only then consider changing the tooling.
Frequently asked questions
How much does AI customer support cost?
Typically a base platform fee plus about $0.99 per resolved ticket, or a per-seat plan for simpler tools. The outcome model is the most defensible for both sides.
What percentage of tickets can AI resolve?
A well-configured agent answers from your own knowledge base and resolves 70–85% of tier-1 tickets; the rest are handed to humans with a summary.
Will AI customer support upset customers?
Not when it answers accurately and escalates cleanly. Frustration comes from bots that improvise — retrieval from your real help centre, with citations, is what keeps quality high.
Want the full playbook?
The free 2026 AI Revenue Playbook turns these shifts into a plan with costs, scripts and pricing templates.