Service order
Service choice → document photo → price and payment → receipt check → manager approves → customer gets the timeline.
Service companies
Scenarios
Scenarios that take the order from "hello" to "paid"
A scenario is a chat flow with steps: ask, collect, verify, confirm. The AI fills in the blanks, checks photos and receipts, and a manager approves with one tap.
- Free plan, no card
- Uzbek, Russian, English
- inPAY and MirPay
- One REST API
Orders in chat are a checklist nobody keeps
Taking an order in Telegram means asking the same five questions every time: which service, full name, passport photo, payment, delivery address. Managers forget one, customers send them out of order, and a receipt screenshot from last week gets attached to today's order.
Button bots can ask questions in order, but they cannot tell a passport from a selfie or read the amount on a receipt.
A scenario knows what to ask and how to check it
You describe the steps: a choice ("which document?"), a collect step ("send a passport photo") with an AI check, a payment step that verifies the receipt against the price, and a confirm step for the manager. ChatVendor runs it in the customer's language, keeps the data, and stops when something does not match.
Scenarios are conversational: if the customer asks a question mid-flow, the assistant answers and returns to the step. If they go off-script repeatedly, the run is escalated to a human with a summary.
- Step types — message, choice, collect (text/photo/document), ai_check, payment, confirm, webhook, jump.
- AI verification — Passport, receipt, product photo — validated with JSON schemas you define.
- Manager approval — Approve, reject or cancel from the panel, a notification, or the API.
- Reminders — A run that stalls gets a polite reminder and then times out.
Building a scenario
-
01
Describe the flow
Use the visual editor or JSON: steps, prompts in three languages, what to save from each step.
-
02
Add checks
Attach an AI check to a step: "is this a passport photo? extract full name and number". Add a receipt check with the expected amount.
-
03
Trigger it
Start on a keyword, on lead intent, or when the assistant decides the customer wants that service.
Scenario features
Visual editor
Steps as cards, branches by choice, jumps and retries — no code.
Document collection
Photos and files are saved with the run and available to managers and via signed URLs.
Receipt verification
Amount, date, card and bank read from the screenshot and compared to the order; wrong currency is rejected politely.
Confirmation step
A manager sees the collected data and approves; the customer gets the outcome message.
Webhook step
Send the collected data to your system mid-flow and use the response in later steps.
Timeouts and reminders
Configurable per step; stale runs are cancelled with a message.
Scenarios in use
Shop order
Product → quantity → address → payment method → confirmation, with the lead updated.
Retail
Registration
Collect name, phone and a photo of an ID for an event or a course; export the list.
Education, events
Scenarios — questions
Do I need a developer?
No. The editor covers the usual flows. Developers can use the JSON format and the webhook step for anything custom.
What if the customer asks something else mid-scenario?
The assistant answers the question and returns to the current step. Repeated off-topic messages escalate the run to a manager.
How accurate is receipt verification?
Vision models read standard bank app screenshots (Click, Payme, Uzum, bank apps) reliably; anything uncertain is sent to the manager instead of being auto-approved.
Can the manager approve from the phone?
Yes — from the ChatVendor panel on a phone, or directly from the notification.
Is the data available via API?
Yes: GET /api/v1/scenario-runs lists runs with collected data, and scenario.done / scenario.failed webhooks fire on completion.