Simple building blocks for precise conversation control
Not every step in an AI agent needs a full AI node.
Sometimes the flow only needs to say something specific.
The Message node sends a fixed message to the user exactly as written.
There is no AI interpretation, rewriting, retrieval, or generation.
If the message configured in the node is:
Thanks. I have everything I need.
that is the message the user receives.
This makes the Message node useful when wording needs to be predictable and controlled.
The Question node asks a single question.
Once the answer is received, the flow continues to whatever node is connected next.
The conversation pauses until the user replies.
The conversation then moves to the next step.
This makes Question useful when the workflow only needs one piece of information and a full AI Form would be unnecessary.
Sometimes it only needs to ask one question, wait for the answer, and continue.
That is where the Message and Question nodes are useful.
They are lightweight building blocks that can be placed at the beginning, middle, or end of a flow and combined with Router, AI Answer, Document Search, Data Search, AI Form, Human Handoff, and other nodes.
They give customers direct control over individual moments in the conversation without adding unnecessary AI behavior.
Question can optionally validate the user's response.
Turn on Validate reply when the answer needs to match a specific format or approved value.
Email validation can be made more specific with optional domain rules.
Customers can define:
The conversation pauses until the user replies.
If the reply does not pass validation:
If any response is acceptable, validation can stay off.
Question can validate several common input types.
Checks that the answer is a valid email address.
Example:
The flow continues only after a valid email is provided.
Checks that the reply is formatted as a phone number.
Useful for:
Checks that the answer is a valid web address.
Example:
What is your company website?
This is useful when the workflow needs a website or online resource before moving forward.
Checks whether the answer matches an allowed value from a list or CSV source.
Examples include:
For example:
What is your store ID?
The user can only continue if the supplied value exists in the configured lookup data.
This is useful when the business wants to confirm that an identifier or option is valid before continuing.
Custom Pattern allows the customer to define a specific format.
For example:
^[A-Z]{2}\d{4}$
could be used for a code consisting of two letters followed by four numbers.
This is useful for structured identifiers such as:
Use it when the expected format is precise and predictable.
Customers can define what the user sees when validation fails.
For example:
Please provide a valid email address.
Or:
Please enter your company email rather than a personal email address.
The message is shown when the answer fails validation, then the Question node asks again.
This keeps the user inside the flow instead of letting invalid data move forward.
AI Form is built for structured multi-question information capture.
Question is built for one answer.
Use Question when you need something like:
For example:
detects:
Existing order
↓
Question
What is your order number?
↓
continues using that information
The node does one job and then gets out of the way.
These nodes are not limited to one part of the conversation.
They can be used at the beginning.
They can connect larger workflow stages.
They can appear before or after AI nodes.
They can be used near the end.
They can also be combined with each other.
For example:
Start
↓
Question
What is your order number?
↓
End conversation
The customer decides how simple or complex the flow needs to be.
Question
What is your order number?
Validation can ensure the number matches the expected format.
↓
handles the request using the retrieved information
This avoids using a multi-question form when one identifier is enough.
Question
What is the best email address for follow-up?
Email validation is enabled.
↓
Message
Thanks. I'm connecting you with our support team now.
↓
This gives the customer precise control over what happens immediately before the transfer.
After a workflow is complete:
Message
Your request has been submitted. Thank you.
↓
End conversation
The closing language is fixed, so every user receives the same final message.
The difference is mainly scope.
You need one answer.
Examples:
You need a structured sequence of answers.
Examples:
Question is better when adding all of that would be unnecessary.
These nodes are often used together, but they do opposite things.
Message
The business sends information to the user.
Question
The business asks for information from the user.
For example:
Message
I found your order.
↓
Question
What email address was used for the purchase?
↓
continues
That simple combination can create a very controlled experience.
Message and Question are intentionally simple.
They are useful because not every step needs AI generation, search, or a full information-capture workflow.
Use Message when the agent needs to say something exact.
Use Question when the agent needs to ask for one thing.
Then connect them to the next node that should handle the conversation.
Together, they give customers precise control over the small moments that connect the larger parts of an AskHandle agent.
