What will AI support cost you?
Per-resolution AI agents typically charge around a dollar or more for each conversation they resolve. On your own AI key, a Dispatch conversation costs what the model costs: about 3.7¢ to 11¢ on the models below. Put in your numbers.
Bring your own AI key and pay your provider at cost — no per-resolution fees. Dispatch plan pricing: see dispatch.osroc.io.
Your numbers
Per-resolution tools bill each conversation their AI resolves. Dispatch on your own key costs the model's price for every conversation. Change the inputs to compare.
| Option | How it bills | Est. AI cost / month |
|---|---|---|
| Dispatch on your key | Claude Haiku 5.5: ≈$0.037 × every conversation | $74.40 |
| Gorgias AI Agent | $0.90 per automated interaction × resolved conversations Plus: A Gorgias helpdesk plan (from $10/month Starter; Basic $50/month annual or $60 monthly) | $1,080 |
| Intercom Fin | $0.99 per outcome × resolved conversations Plus: Intercom seats (from $29/seat/month), or standalone on your current helpdesk | $1,188 |
| Redo AI Support Agent | $0.99 per resolution × resolved conversations | $1,188 |
Prices checked 17 October 2026, from vendor pricing pages (linked on each comparison). Vendor figures use their published per-resolution list price and leave out required base plans, annual discounts and volume deals, so they are a floor, not a quote. Dispatch's own plan fee is left out too, just as vendor base plans are. Check vendor sites for current pricing.
At common volumes
The same comparison at four volumes, assuming the AI resolves 60% of conversations.
| AI conversations / month | Gorgias AI Agent $0.90 per automated interaction | Intercom Fin $0.99 per outcome | Redo AI Support Agent $0.99 per resolution |
Dispatch on your key Claude Haiku 5.5 to Claude Sonnet 5.5 |
|---|---|---|---|---|
| 500 300 resolved by AI |
$270 | $297 | $297 | $18.60 – $55.80 |
| 2,000 1,200 resolved by AI |
$1,080 | $1,188 | $1,188 | $74.40 – $223 |
| 5,000 3,000 resolved by AI |
$2,700 | $2,970 | $2,970 | $186 – $558 |
| 10,000 6,000 resolved by AI |
$5,400 | $5,940 | $5,940 | $372 – $1,116 |
The math behind Dispatch's number
With your own key, a conversation costs whatever your AI provider charges for the tokens it uses. No markup, no per-resolution fee.
| Model (your key) | Provider | List price per 1M tokens (in / cached / out) | ≈ per conversation |
|---|---|---|---|
| Claude Haiku 5.5 | Anthropic | $1 / $0.1 / $5 | $0.037 |
| Claude Sonnet 5.5 | Anthropic | $3 / $0.3 / $15 | $0.11 |
How we estimate: a typical conversation is about 4 shopper messages. Each message takes a few model calls (reading your catalog, policies or an order), adding up to roughly 60,000 input and 3,000 output tokens per conversation. Where the provider offers prompt caching, we assume 70% of input is served from cache. These are design estimates, not measured averages: long conversations cost more, short ones less. Prices are each provider's list price as recorded in Dispatch's model catalog. We only show models Dispatch has certified; cheaper models from OpenAI, DigitalOcean and OpenRouter can be used on your own key but aren't certified yet, so they aren't counted here.
Claude Haiku 5.5 list price: $1/M input, $0.1/M cached input, $5/M output fresh input 18,000 tokens × $1/M = $0.0180 cached input 42,000 tokens × $0.1/M = $0.0042 output 3,000 tokens × $5/M = $0.0150 per conversation ≈ $0.037
Keeping the comparison fair
What per-resolution pricing gets right
You only pay when the AI finishes the job, and the vendor carries the model cost and tuning. For low volumes, a resolution fee can be simpler than managing your own AI key.
What it leaves out
A required helpdesk or chat plan, how "resolved" is defined, and that the bill rises as automation improves. With your own key, cost tracks the model, and handed-off conversations cost only the model calls made before the handoff.
Questions
What does “bring your own key” mean?
You create an API key with an AI provider (Anthropic, OpenAI, OpenRouter or DigitalOcean) and paste it into Dispatch. Conversations run on your provider account at their list prices, so you see the real cost on your provider's invoice. Keys are checked when you save them, encrypted at rest and never shown again.
Why is my own key so much cheaper than a resolution fee?
A resolution fee bundles the model, the vendor's platform and margin into one price per outcome. The model part alone is usually a small fraction of that. With your own key you pay the model part directly, plus a Dispatch plan (pricing: see dispatch.osroc.io).
Are these estimates exact?
No. Vendor figures use published list prices and leave out required base plans and any discounts. Dispatch figures use our estimate of tokens in a typical conversation; longer conversations cost more, shorter ones less. Your provider's dashboard shows the real number once you're live.
What does a conversation handed to my team cost?
On your own key, only the model calls made before the handoff. Your helpdesk's own pricing still applies to the tickets it receives.
Do I need my own key?
No. Dispatch can also run on models we host for you; see dispatch.osroc.io for plans. Add your own key whenever you want to pay your provider directly.
Keep reading
The AI front door to your helpdesk
Answers shoppers instantly, sells while it helps, and hands the hard ones to your team in the tools they already use.