Create a workflow using AI for automating personalized report generation

Help me Build my Workflow
1

I need help building an AI workflow using Gemini. The process involves:

  1. Creating a database using a survey carried out with sellers.
  2. My team imports data (bonus payments) into a spreadsheet.
  3. The AI reads this information.
  4. The AI generates a personalized report, creating market intelligence for our clients by connecting the survey data with the bonus payment details.

We are restricted to using Gemini, so any guidance on how to develop this workflow—from APIs to prompt engineering—is welcome. I’m looking for guidance on how to best implement this workflow and leverage Gemini’s capabilities for this task.

2

Hey @tamires_vianna hope all is good. Welcome to the community.

If you would like to explain in more detail:

  • do you intend the database ingestion to be a part of the n8n workflow somehow?
  • same goes for the bonus payments ingestion to the spreadsheet
  • could you provide examples of
    • spreadsheet rows
    • generate report
3

Hey, @jabbson. Thanks for replying.
About your questions:

  • I did some research and found some interesting templates using RAG - i think the database ingestion can be created using something like that. For the bonus payments ingestion, I’m not sure the best way to import, but the spreadsheet contain the rows: name, document, total bonus payment (R$)
  • I don’t have a defined report just yet, but it’d be nice to have the name of the company that made the payments, the colaborator responsible for the transaction and the insights
4

Ok, this sounds good. Could you give me a rundown of what you think the flow should do, in terms you are most comfortable with. Just your vision of possible action flow including input data, data transformation if needed, as well as the output data.

5

I need a workflow that allows my team to import spreadsheets via Slack or Notion. Once imported, we’ll name the company associated with the data so the insights are relevant to our CS team.

Here’s the vision for the workflow:

  1. Input Data: My team will upload a spreadsheet containing bonus payments. This can be done directly through Slack or Notion. The input will also include the name of the company the data belongs to.
  2. Data Transformation: The AI, which has been pre-trained on a database from a salesperson survey, will analyze the spreadsheet. It will use its knowledge from the survey data to generate meaningful insights and a performance report.
  3. Output Data: The AI will create a comprehensive report based on the insights. This report will then be made available and attached to the company’s card in HubSpot.
6

(post deleted by author)

7

The first part is straightforward, ingesting documents is an easy part.

For the transformation and output… you previously mentioned RAG as possible solution. Did you research what it is and what problem does it solve? Basically it helps with creating and then retrieving data from a database. You can retrieve data from it based on semantic similarities with the input text. In simple terms - you say “cars” and it goes and searches if there were any parts of the pre-trained corpus that was about cars and returns it. Does this sound like something that would solve your transformation portion of the flow?