Skip to main content

Voice to quote request

This workflow turns a spoken note into a quote request in Propeller. A sales rep records after a call who the customer is and what they want, and a request with that customer and those products is ready for review.

The recorder is not included

The workflow receives a recording and does the rest. The app or form with which a rep records and sends it is not part of the example. A technical colleague has to build that part.

In short
Where you use itIn the background. The results show in the Action Hub and under Sales > Requests
StartsWhen a recording arrives at the address of the workflow
Changes in PropellerCreates quote requests and actions
Uses AIYes, OpenAI. It gets the recording and the names of customers and products
TextsEnglish and Dutch
Runs in n8nOne per recording
DifficultyAdvanced. You build the recorder yourself

What you get​

A quote request with the customer, the products, the quantities and the remarks from the note. An action in the Action Hub opens it and names the products that could not be matched.

An action in the Action Hub for a quote request made from a voice memo, with 2 product lines and the View order button

When no request can be made, the action "Voice memo needs manual review" says why and shows the start of what was said.

What you need​

  • An n8n account with the credential Propeller API key. See Set up n8n.
  • An API key with Owner access on User management and Order Data and Viewer access on Product Data.
  • An OpenAI account. You pay OpenAI per use.

How it works​

  1. AI writes out the recording and picks out the customer, the remarks and the products.
  2. The workflow searches companies and contacts by the names in the note. The AI picks from at most 20.
  3. Each product is matched on an article number that was spoken. Without one the workflow searches the catalogue and the AI picks from at most 10 products. Quantities always come from the note.
  4. The workflow creates the quote request and the action. Propeller calculates the prices for that customer.
Good to know

The workflows and agents in the Agent Hub are free to use and adapt. They are examples, not supported product features. Test them in a test environment and check the results before you use them on real customers.

Install it​

Install a workflow from the library explains each step.

  1. Import the file into n8n.
  2. Pick your OpenAI account on the four AI nodes: Transcribe Recording, Extract Request Data, Match Customer and Match Products.
  3. Check that the eight nodes that talk to Propeller use the credential Propeller API key.
  4. Open the Settings node and set the values of the table below.
  5. Open the Webhook node and create a login with a username and a password you make up.
  6. Copy the Production URL of the Webhook node.
  7. Click Publish. The address only answers after that.
  8. Set up your recorder to send the sound file to that address, in a field called audio, with the same username and password.
  9. Record a test note that names a customer and a product.

After the test you find a new request under Sales > Requests and an action in the Action Hub.

Make it yours​

SettingDefaultWhat it does
CHANNEL_ID1The channel the request is created in. The number is in the Id column under Shops & Channels > Channels
ASSIGN_TO_ADMIN_USER_IDnullThe colleague who gets the actions. The number is in the Id column under Admin > Backoffice Users
LANGUAGE'NL'The language of the product names that is used to find products
TRANSCRIBE_LANGUAGE'nl'The language your reps speak in
ORDER_STATUS'REQUEST'The status the new request gets

Limits​

  • Keep a note short. A long recording can be too large for n8n or for the service that writes it out.
  • Names that sound alike can lead to the wrong customer or product. Always review the request.