Billing-event sandbox
Twelve conversations your AI will have. Move the sliders to match your month. Each vendor gets a stamp per conversation and a bill range. Click a stamp for the excerpt, URL and viewed date.
Shares are weights; they always add to 100%.
The AI answers and the customer says thanks or that it solved the problem.
The AI answers. The customer never replies and gives no feedback.
After the AI answer, the customer asks for more help or for a person.
The AI cannot answer and routes to a person under the general escalation rules.
The AI runs a workflow you configured (check an order, collect details) and then hands to a person.
A teammate replies on their own while the AI conversation is open.
The customer asks a second question in the same conversation and the AI answers it.
The same customer comes back about the same problem after a first AI answer.
The AI gives a wrong or unhelpful answer and the customer silently gives up.
The AI checks an order or runs an action and reports the result, with no confirmation.
A message that is not a real customer request but the AI replies to it.
The customer asks three separate questions and the AI answers all three.
BILLED the vendor's text says it bills · NOT BILLED says it does not · VERIFY text is ambiguous, conflicts, or a vendor-side check decides · DOCS SILENT text we read does not say
| Vendor | Month, low to high | S1 | S2 | S3 | S4 | S5 | S6 | S7 | S8 | S9 | S10 | S11 | S12 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Fin (Intercom)Outcome | $3,367 to $3,614 | BILLED | BILLED | NOT BILLED | NOT BILLED | BILLED | BILLED | BILLED | DOCS SILENT | BILLED | BILLED | NOT BILLED | BILLED |
| Zendesk AI agentsVerified automated resolution | $0 to $7,002 | VERIFY | VERIFY | NOT BILLED | NOT BILLED | NOT BILLED | NOT BILLED | VERIFY | DOCS SILENT | VERIFY | VERIFY | DOCS SILENT | DOCS SILENT |
| Help Scout (Envoy / AI Answers)Resolution | $1,613 to $2,401 | BILLED | BILLED | NOT BILLED | NOT BILLED | NOT BILLED | NOT BILLED | NOT BILLED | DOCS SILENT | BILLED | DOCS SILENT | DOCS SILENT | VERIFY |
| Gorgias AI AgentAutomated interaction | $990 to $3,951 | BILLED | VERIFY | DOCS SILENT | NOT BILLED | DOCS SILENT | NOT BILLED | DOCS SILENT | DOCS SILENT | VERIFY | DOCS SILENT | NOT BILLED | DOCS SILENT |
| Tidio (Lyro)Lyro conversation | $3,501 to $3,901 | BILLED | BILLED | BILLED | BILLED | BILLED | BILLED | BILLED | BILLED | BILLED | BILLED | BILLED | BILLED |
| Crisp (Hugo)AI usage credits per conversation | $250 to $300 | BILLED | BILLED | BILLED | BILLED | BILLED | BILLED | BILLED | BILLED | BILLED | BILLED | BILLED | BILLED |
| LiveChat (Text) AI AgentResolution | $1,634 to $4,357 | BILLED | DOCS SILENT | DOCS SILENT | NOT BILLED | DOCS SILENT | DOCS SILENT | BILLED | DOCS SILENT | DOCS SILENT | DOCS SILENT | DOCS SILENT | BILLED |
| Freshworks Freddy AI AgentSession | $2,205 to $2,450 | BILLED | BILLED | BILLED | BILLED | BILLED | BILLED | BILLED | VERIFY | BILLED | BILLED | VERIFY | BILLED |
| HubSpot Breeze Customer AgentResolved conversation (50 credits) | $550 to $1,751 | BILLED | VERIFY | NOT BILLED | NOT BILLED | NOT BILLED | NOT BILLED | VERIFY | DOCS SILENT | VERIFY | DOCS SILENT | DOCS SILENT | DOCS SILENT |
| Salesforce AgentforceConversation (24-hour session) or action | $9,002 to $10,002 | BILLED | BILLED | BILLED | BILLED | BILLED | BILLED | BILLED | VERIFY | BILLED | BILLED | DOCS SILENT | BILLED |
| AdaConversation (custom contract) | no public unit price | VERIFY | VERIFY | VERIFY | VERIFY | VERIFY | DOCS SILENT | VERIFY | DOCS SILENT | VERIFY | VERIFY | DOCS SILENT | VERIFY |
| eesel AICredit (one per ticket or chat) | $1,599 to $1,750 | BILLED | BILLED | BILLED | BILLED | BILLED | VERIFY | BILLED | VERIFY | BILLED | BILLED | VERIFY | BILLED |
| HelplyAI credit per action | no public unit price | BILLED | BILLED | BILLED | BILLED | BILLED | VERIFY | BILLED | BILLED | BILLED | BILLED | NOT BILLED | BILLED |
| TalvoResolution | $299 to $1,034 | BILLED | DOCS SILENT | NOT BILLED | NOT BILLED | NOT BILLED | NOT BILLED | BILLED | DOCS SILENT | DOCS SILENT | DOCS SILENT | DOCS SILENT | BILLED |
| VarleyResolution | $550 to $1,225 | BILLED | NOT BILLED | NOT BILLED | NOT BILLED | NOT BILLED | NOT BILLED | DOCS SILENT | DOCS SILENT | NOT BILLED | VERIFY | DOCS SILENT | DOCS SILENT |
| ChatbaseMessage credit per AI reply | no public unit price | BILLED | BILLED | BILLED | BILLED | BILLED | BILLED | BILLED | BILLED | BILLED | BILLED | BILLED | BILLED |
Low counts only BILLED stamps. High also counts VERIFY and DOCS SILENT, at the high end of the price range. Prices that sources disagree on become a range (see each vendor's file). One AI conversation is one billable event at most. Plan fees, seats, taxes and allowances are excluded, except eesel and Talvo whose plans are the unit. Ada, Helply and Chatbase have no per-conversation unit price in our sources, so they have stamps but no bill.
Scenarios
- S1 Answered, customer confirms it worked: The AI answers and the customer says thanks or that it solved the problem.
- S2 Answered, then 24 hours of silence: The AI answers. The customer never replies and gives no feedback.
- S3 Customer says "I still need help": After the AI answer, the customer asks for more help or for a person.
- S4 Escalated by rule, human resolves: The AI cannot answer and routes to a person under the general escalation rules.
- S5 Procedure ends in a hand-off: The AI runs a workflow you configured (check an order, collect details) and then hands to a person.
- S6 Human steps in before the customer asks: A teammate replies on their own while the AI conversation is open.
- S7 Follow-up question, AI answers again: The customer asks a second question in the same conversation and the AI answers it.
- S8 Same issue returns within 72 hours: The same customer comes back about the same problem after a first AI answer.
- S9 Answer was wrong, customer leaves: The AI gives a wrong or unhelpful answer and the customer silently gives up.
- S10 AI only looks up or acts: The AI checks an order or runs an action and reports the result, with no confirmation.
- S11 Spam, test or bot message: A message that is not a real customer request but the AI replies to it.
- S12 Three questions in one chat: The customer asks three separate questions and the AI answers all three.