What AI Support Should Actually Cost a Small Team

The pricing model is the product decision

When you're evaluating an AI support agent, the feature list matters less than the pricing structure. Features determine what the agent can do; pricing determines whether you can actually let it run.

Intercom Fin charges per resolution — as of their published pricing page, that's $0.99 per resolved conversation. That number feels low in isolation. It isn't, once your agent starts doing its job.

Say a team resolves 800 tickets a month and their AI handles 60% of them. That's 480 AI resolutions at $0.99 each — roughly $475 per month, on top of whatever Intercom seat licenses cost. If the agent gets better — say it reaches 75% containment — the monthly AI bill climbs to around $594. You're paying more because your tooling improved. That's a strange incentive structure.

Chattering charges a flat monthly rate. The agent can resolve one ticket or a thousand; the invoice doesn't move. For a small team trying to set a support budget before a board meeting, that distinction is material.

What 'per resolution' actually measures

It's worth being precise about what Fin counts as a resolution. Per Intercom's published documentation, a conversation is resolved when the customer confirms their question was answered, or when the session ends without escalation. That definition isn't unreasonable, but it means your costs are tied to outcomes — which are exactly what you want the AI to produce.

The perverse result: if you optimize Fin's training, tune its knowledge base, and push containment rates higher, your monthly spend climbs. The vendor's upside is directly coupled to your agent's success. That might be acceptable for an enterprise with a predictable, high-volume support operation. For a 10-person SaaS team with seasonal spikes, a product launch, or a viral bug report, it creates real budget variance.

Customers increasingly expect a reply within the hour, not the next business day. Small teams deploy AI support agents precisely because they can't staff for that expectation. The moment the agent works well enough to meet it, costs spike.

What Chattering includes at a flat rate

Chattering's agent is embedded directly in your website or docs, answers questions by crawling and citing your actual content, and hands off to a human with full conversation context when it can't resolve something. That handoff isn't a dead end — the conversation and context land in a shared inbox, so whoever picks it up doesn't start cold.

The flat monthly price includes the agent, the shared inbox, a self-hosted help center, and a public feedback portal. There's no per-seat fee layered on top, no resolution counter running in the background, and no surprise charge when a product update drives a spike in tickets.

For comparison, Intercom's Fin is an add-on to the core Intercom platform. You're paying Fin's per-resolution fee on top of existing Intercom seat licensing — which, per their published pricing, starts at $29 per seat per month on the Starter plan, scaling significantly from there for features like custom roles, SLAs, or reporting.

The operational reality for a small team

Small SaaS support teams make decisions differently than enterprise procurement teams. A VP of Support at a 200-person company can absorb a variable AI line item because it's a rounding error in a larger budget. A founder running support for a 15-person company needs to know, in January, what support will cost in June.

There's also a training investment to consider. Whichever tool you deploy, someone needs to:

• Connect your docs, help center, and knowledge base
• Review what the AI gets wrong in the first 30 days
• Tune escalation triggers and fallback messages
• Monitor resolution quality, not just resolution rate

That work takes real hours. If the pricing model punishes you for doing it well, the incentive to invest in that tuning erodes.

With flat-rate pricing, the calculus changes. A better-trained agent means lower human workload, faster response times, and a better customer experience — with no corresponding bill increase. The team captures the full upside of the investment.

When Fin is the right answer anyway

Fin is a reasonable choice if you're already deep in the Intercom ecosystem — you have CRM data, conversation history, and workflows built on top of it. Switching has a real cost, and per-resolution pricing may be acceptable if your ticket volume is stable and low enough that the math stays predictable.

It's also worth noting that Fin has a longer market track record and a larger enterprise reference base. If your sales process requires that kind of vendor credibility, it matters.

But if you're evaluating fresh — no existing Intercom contract, no deep workflow dependency — the pricing model alone is a meaningful reason to look at alternatives. Paying a flat rate for infrastructure that includes your AI agent, inbox, help center, and feedback loop is a different category of product decision than licensing a per-outcome AI add-on.

The question to ask before you sign

Before committing to any AI support agent, run this scenario: what does the monthly cost look like if containment rate doubles? If the answer is 'it also doubles,' make sure your budget can absorb that before the contract renews.