The AI Prompt action allows your workflow to use AI to generate or transform text based on earlier steps. It helps you to summarize, reword, or interpret incoming data.
This action is especially useful when you need the workflow to “understand” or “reword” something, rather than just extract or pass it along.
💡Tip
We recommend using the AI Prompt to extract data from non-standardised emails / incoming communication from clients.
When to use it
Examples of when you might want to use the AI Prompt node are to:
- Summarise an email before sending it to a CRM or ticketing system
- Extract the customer’s main request from a long body of text which is not standardized
Configuration
There are no mandatory fields in an AI Prompt action node.
Prompt field
Write your instruction to the AI as a prompt in the Prompt field, for example:
“Summarise the following email in two bullet points”
“What is the customer asking for in this message?”
Tips for writing successful prompts
- If you want the AI to use data from a previous step in the workflow, (like the body of the email which started the workflow), click + Add a variable and choose the relevant variable. This allows your prompt to dynamically reference data from previous nodes.
For example, your full prompt might be:
Extract what is the customer's main request from #textBody
- If you want to give the AI some examples of how to do something, like to classify data, give it the examples and instructions first before putting the information you want it to review.
Here is an example where we set out the terms for the AI to look for before we give it the variable of #attachmentsbody to check:
Creativity slider
Adjust how the AI responds to your prompt. Move towards Deterministic & Focused for precise, predictable results, or towards Creative & Unpredictable for more varied, imaginative outputs.
Output values
Give your output a name (this becomes the variable name you can use later in the workflow) and choose the type of data it will contain, such as text, number, email, phone, etc.
Output
The AI’s response is saved as an output value, which appears as a custom variable you can use later in the workflow. It will be listed with the variable name you defined in the Output values section.
You can then use the output further down in the workflow. For example, to Create a Job in BigChange, send in an email etc.
Example use case - Summarising job requests for job descriptions using AI
Smith & Co want to streamline how customer repair requests are turned into clear, ready-to-use job descriptions in BigChange.
Some of their smaller clients do not send job requests via the job request form but rather they handtype their request to a job request email address.
While they use the Extract Text action to extract the information about the job being requested from requests which come from the auto-generated emails, an AI Prompt action will work best here to extract the relevant data from the handtyped emails.
They will configure the AI prompt action with the instruction:
"Summarise the customer’s request in one concise sentence suitable for a job description."
They will choose the Body as the source of the data to extract so the AI reviews the customer’s original message.
The AI-generated output will appear as a custom variable, which they then map into the Job Description field of a Create Job in BigChange node.
This is a great solution for automating job creation in BigChange or any other job scheduling tool, where freehand emails are sent to a specific email address and are not auto-generated.