How does information move between nodes?

Information collected earlier in a conversation can be used later in the workflow.

This helps the AI agent avoid asking the customer for the same information again.

Examples include:

• Name
• Email
• Phone number
• Location
• Order number
• Product preference
• Service type
• Appointment preference
• Values used in a calculation

The exact information available depends on what the workflow has collected and how the nodes are configured.

Why this matters

A good AI agent should feel like one continuous conversation.

For example, if the customer already said:

“My name is Sarah and I need service in Boston.”

the workflow should not ask for the name and location again unless there is a reason to verify them.

Information can come from the conversation

AI Answer can use information the customer has already provided conversationally.

For example:

Customer:

“I need a quote for a 2,000 sq ft property.”

If the calculation needs property size, the value may already be available.

The AI agent should not ask for it again unnecessarily.

Question saves one specific value

Use Question when you need one controlled answer.

Examples:

• Email address
• Order number
• Engine power
• Location

The saved answer can be used later in the workflow.

AI Form saves several values

Use AI Form when you need multiple structured answers.

For example:

• Name
• Email
• Service needed
• Location
• Preferred date

Each answer is saved before the form moves to the next question.

These values can then support later steps.

Example: AI Form to Human Handoff

A support workflow might use:

AI Form → Human Handoff

The AI Form collects:

• Customer name
• Email
• Order number
• Issue details

The Human Handoff then receives the conversation after the important information has already been collected.

This gives your team useful context immediately.

Example: Question to Calculation

A workflow might ask:

“What is your vehicle's engine power in kW?”

The Question node saves the answer.

Later, AI Answer with Calculations enabled can use that value in the formula.

Example: AI Form to Scheduling

An AI Form could collect:

• Name
• Email
• Service type

Then AI Answer with Scheduling enabled can continue the booking process without asking for the same contact information again if it is already available.

Example: Router after collected information

Information collected earlier can also affect the next step.

For example, a workflow might collect a customer's location first.

Then the next part of the workflow can use that information to determine which branch, service, or team is appropriate.

Keep field meaning consistent

If a saved value represents one thing, keep that meaning consistent.

For example:

location

should not mean:

• Customer city in one path
• Store location in another path
• Country in another path

Use clear field names so later nodes know what the value represents.

Use clear names for saved values

Good examples include:

• customer_email
• order_number
• service_location
• engine_power_kw
• preferred_date

Avoid vague names such as:

• value1
• answer
• data
• info

Clear names make workflows easier to maintain.

Do not collect information earlier than necessary

Just because a value can be reused does not mean you should collect it immediately.

For example, do not ask for an email address at the start of every conversation if it is only needed for one specific action later.

Collect information when it becomes relevant.

Avoid duplicate questions

If the customer already provided a value, the workflow should not ask for it again unless:

• The value was unclear
• Verification is required
• The customer changed the answer
• The business process requires confirmation

Repeated questions create unnecessary friction.

Use AI Verification when a value must meet a requirement

AI Form can use AI Verification to check whether an answer is acceptable.

For example:

• A valid order number format
• A supported location
• A required type of response

If verification fails, the AI Form can re-ask before saving the value.

Use AI Answer Cleanup when the saved value should be normalized

AI Form can use AI Answer Cleanup to store a cleaner value.

For example, the customer says:

“I think it's around one hundred fifty kilowatts.”

The saved value might need to become:

150

This is useful when later nodes need consistent input.

Let the customer correct information

If the workflow includes confirmation, customers can review collected information and correct a value when needed.

This is useful before:

• Submitting a request
• Running a calculation
• Booking an appointment
• Handing the case to a person

Do not assume every node needs every value

Only pass or use information that is relevant to the next step.

For example, a Document Search question may not need the customer's phone number.

Keeping the workflow focused reduces unnecessary handling of data.

Test information reuse

During testing, provide values early in the conversation and confirm that later nodes use them correctly.

Try:

• Providing a name before the form
• Giving a location in the first message
• Changing an answer later
• Giving several values in one message
• Leaving one required value missing

Check whether the workflow asks only for what is still needed.

Test corrected values

If the customer changes information, confirm that the updated value is used.

For example:

“Actually, use my work email instead.”

or:

“I said Boston, but the service is in Cambridge.”

The later workflow should use the corrected information.

Tips for moving information between nodes

• Collect information only when it is needed.
• Use Question for one value and AI Form for several.
• Give saved values clear names.
• Reuse information the customer already provided.
• Avoid asking the same question twice.
• Verify important values before using them.
• Normalize values when later steps require consistent formatting.
• Let customers correct information when necessary.
• Test whether later nodes use the latest value.

Related guides

For more information, see:

• What is an AI Form node?
• What are Message and Question nodes?
• How do I plan an AI agent?
• How do Calculations work?
• How does Scheduling work?
• How does Human Handoff work?