Bot workflows
Scripted question flows in the Aimogen Pro chatbot, where the bot asks predefined questions and branches on the answers.
A bot workflow scripts part of the conversation: the bot asks defined questions in order, branching on the answers, before or instead of free-form chat.
AI Chatbot › Bot Workflows
Why script a conversation#
Free-form chat is right when the visitor knows what to ask. A workflow is right when you know what to ask:
- Qualifying a lead before handing to sales
- Structured triage before a support handoff
- A booking or quote flow needing specific fields
- Onboarding
The bot asks the questions in a known order, so you get complete, comparable answers rather than whatever the visitor thought to mention.
Configuration#
The flow is stored as a configuration string in Bot Workflow Configuration String, built in the workflow
editor.
Each step defines:
| Property | Meaning |
|---|---|
| The question | What the bot asks |
| Expected answer type | Free text, a choice, or a validated value |
| Branching | Which step comes next, based on the answer |
| Storage | Where the answer is kept for later use |
Per placement:
[aimogen-chat-form bot_workflow="lead-qualification"]Designing a flow#
Keep it short. Every question is a chance for the visitor to leave. Four or five is usually the practical limit.
Ask the qualifying question first. If the answer to "are you an existing customer?" changes everything after it, ask it first.
Allow escape. Someone with an unusual question should be able to reach free-form chat rather than being trapped in a script.
Confirm at the end. Repeat what was captured before submitting.
A lead flow:
1. "What are you looking for help with?"
-> pricing -> step 2
-> technical -> step 4
-> something else -> free chat
2. "How many sites do you manage?"
3. "What is the best email to send pricing to?" -> capture -> end
4. "Which plugin version are you on?"
5. "What is happening, and what did you expect?" -> capture -> handoffCombining with free chat#
A workflow does not have to own the whole conversation. Common patterns:
- Workflow first, then free chat. Qualify, then let the visitor ask anything.
- Free chat, workflow on demand. Free chat normally, with the flow triggered when the visitor asks about something specific.
- Workflow only. A pure form-like interaction. If that is all you need, consider whether an AI Form is a better fit — it is simpler and cheaper.
Using the answers#
| Destination | How |
|---|---|
| Lead record | Combine with lead capture — answers are stored as aiomatic_lead posts |
| An OmniBlock workflow | Trigger the ai_chatbot_omniblock block, which receives the full input |
The email extension, or an OmniBlock send_email block | |
| A webhook | The webhook extension, or an OmniBlock send_webhook block |
| A CRM | Through the webhook, into your automation platform |
Workflows versus OmniBlocks#
They complement each other:
| Bot workflow | OmniBlock | |
|---|---|---|
| Runs | Inside the conversation | Server-side |
| Interacts with the visitor | Yes | No |
| Does work | No | Yes |
| Good at | Collecting structured input | Processing it |
The strong pattern is a bot workflow that collects, then triggers an OmniBlock that acts.
Common problems#
The bot ignores the workflow. The configuration string is not applied to this placement — check the
bot_workflow attribute or the global setting.
Branching goes to the wrong step. The answer did not match any branch condition. Add a fallback branch for unexpected answers.
Visitors abandon part-way. The flow is too long, or a question is unclear. Reduce it, and check the first question is one people can answer instantly.
Related#
Still stuck? Open a support ticket and include the diagnostics from Aimogen Pro › System & Logs › System Info.