top of page
  • Writer's pictureIlia Zelenkin

Extracting Data from Business Cards Part 2 - Detecting Company Industry

All righty! You already have an idea of what can be done with modern tools in a matter of hours. This tech allows you to perform on the next level without hiring an extra headcount. This means a small firm with less than 50 people have access to tools and automations that allows it to perform as 500 people team.


In this article let's explore how we can improve our existing automation. What we will do is:

  • Add an automated industry detection

  • Add an automated voice message that you can send yourself moments after you took a photo of the business card.


Here is the full video:



Adding Industry detection

To add an industry detection we will create a new Bitskout plugin. This will be an Understand plugin type:



We will start adding examples. Examples will allow us to set our own industry definitions and not use some pre-defined ones.


So let's add our examples. You don't need to add all possible examples, start with a few, and then if Bitskout makes a mistake, you can always add it as a new example.



Let's save our plugin and test it:



As you can see that we did not add Walmart in the above list of examples and Bitskout picked it up automatically.


Adding Bitskout plugin to Make.com

Let's add our plugin to the make.com scenario. We will need to choose "Run Plugin for Text" as our analysis will be based on the company name from the previous steps.



Let's run our plugin once to get the field name to make.com. You can do that by right-clicking on the Bitskout module and then selecting "Run this module only" and then specifying the company name manually.



Bitskout will run and return the label.

Now let's map our industry to the monday.com task - in our case, we are using the Dropdown field in monday.com and thus we will directly map the value to it.



So now every time we upload a photo of a business card to the designated folder, the information will be automatically extracted and the industry will be detected as well.



Creating a voice message

Okay, this is one is a stretch - however, a good test of what is possible today. Last time, we sent ourselves a message via Slack (it can be done via WhatsApp or FB Messenger) like this:



Adding industry type to become more focused

As now we extract the industry from the company, we can add more specificity to openAI and have better-focused topics to talk about. To do that we just need to modify the prompt like this:


And now your messages will be more specific. You can add other dimensions to it and modify the prompt even better.

Generating Voice Message

So now let's convert the openAI prompt into a voice message that we can send to ourselves and quietly listen to the earphone.


We will use Google Speech-to-Text Synthesis in this example, but you can choose simpler solutions. The problem is that Google requires a GCloud account and it is tricky to set it up.

So let's add a Google Speech-to-Text module:



The module will generate a file. Now you need to send this file to yourself. We will use the Slack Upload a file action to load the file to our Slack channel:



And now every time you run the scenario, you will get a file with a voice message to give you tips to talk with a person about:



Conclusion

As you can see in a short period of time we managed to create a pretty advanced workflow that can help you run better events, and interviews, and nail the networking. Such automations don't require piles of money or technical expertise.


If you want to deploy it for yourself, here is the scenario blueprint that you can import to make.com. And as the Bitskout plan includes setup calls, you can just start a trial and get the above automation configured for you in an hour.




5 views0 comments
bottom of page