Who wrote this: we make Kerux, one of the tools on this page, so weigh our view of it accordingly. Prices come from each vendor's own pricing page. We did not run hands-on benchmarks of every product, so we won't pretend to rank them on quality. We sort by fit instead.
1. AI is billed by what it does, not who uses it
Intercom meters Fin at $0.99 per outcome. Zendesk bills $1.50 per committed automated resolution, Help Scout's AI Answers $0.75 per resolution, and Gorgias $0.90 or more per interaction. That makes the support budget a forecast of volume, which most finance teams didn't have to make before. Do now: model your busiest month, not your average.
2. Everybody counts something different
Outcome, resolution, session, interaction, conversation. Freshdesk counts a 72-hour email window as a single session, while Gorgias also counts each interaction as a ticket. A lower rate can be a bigger bill. Do now: compare on your real tickets, using each vendor's definition.
3. Seats aren't going away
You might expect AI meters to replace seat prices. On these pages, they sit on top of them. Every vendor here except Gorgias still charges for people. So if AI handles more and your team stays the same size, you pay both. Do now: audit who really needs a paid seat, and ask about read-only access.
4. The agent is separating from the help desk
Intercom's pricing page says Fin can run on other help desks. Front lists External Agents, which let any AI agent join the team. Freshdesk now lists MCP actions as an add-on, and we built a Claude connector ourselves. The direction looks like help desk as the record, and AI as something you can swap. Do now: ask any vendor how open its data and actions are to outside AI.
5. Spending controls are becoming a feature
Once a bill follows volume, someone asks who can stop it. Intercom's pricing page now promotes spending limits and alerts, and Kerux ships a monthly AI spend cap. Do now: before you turn on any AI agent, find out whether you get an alert, a hard stop, or neither.
6. The gap between model cost and meter price is visible
A support answer on a small, fast model costs roughly 3¢ in tokens, though it runs 1 to 5¢ by model and length. Meters also pay for tuning, reliability and support, and that has real value. But buyers can now see the gap. That is why bring-your-own-key options exist, ours included. Do now: see the maths per ticket, then decide whether the extra is worth it for you. For plenty of teams it will be.
What we expect in 2027
This part is our opinion, not data, and we sell a product that benefits if we are right.
- Hand-off quality becomes the main way to compare AI agents, since the easy questions will be handled by everyone.
- More vendors publish rates. Zendesk began listing its rate in 2026, and buyers now expect to see one.
- Help centers get more attention, because every AI answer is only as good as the page behind it.
Ready to shop? Start with the 2027 help desk guide.
Sources: Intercom pricing (checked 2026-11-05) · Zendesk pricing (checked 2026-11-05) · Help Scout pricing (checked 2026-11-05) · Freshdesk pricing (checked 2026-11-05) · Gorgias pricing (checked 2026-11-05) · Front pricing (checked 2026-11-05). Kerux figures from our own pricing page. Vendor pricing changes, so confirm on their page before you decide.
More 2027 buying guides
- Best help desk software for 2027: seven tools at published prices
- Best customer support software for small businesses in 2027
- Best AI customer support software for 2027, and what each one meters
- How much does help desk software cost in 2027?
- Best Zendesk alternatives for 2027
- Best Intercom alternatives for 2027
- Best Help Scout alternatives for 2027
- Best Freshdesk alternatives for 2027
Questions people ask
What is the biggest customer support trend for 2027?
AI billing. Every major help desk in this guide now meters its AI agent somehow, which turns support cost from a headcount number into a volume number.
What is outcome-based pricing in customer support?
You pay when the AI achieves a defined result, such as a resolved conversation, rather than per seat or per message. The important detail is how each vendor defines that result.