Yes, now you can easily build, train, and publish AI models without writing a single line of code using pre-built or configurable templates and a guided process.
The ability is great, truly! Tried it and loved it. You must create some test data for the tool to train from- that can take a while. But after that, defining the AI model was quick and painless.
However, my concern is- what can you actually do with it?
So, I was brainstorming with myself and with my colleges and came up with some ideas. I share them here with you, just to give you a starting point to play with the tool and figure out how your organization can benefit from the AI tool.
My suggestion for you- fake it till you make it. Try creating test data as I describe here and run an AI model. After you try it once you will have a better idea of its usage. That might trigger some usage ideas, for your organization’s benefit.
Here is an idea for a text classification AI model-
Our company receives many inquiries daily, mostly via emails. Different people tend those inquiries, based on the subject at hand. Assigning the emails to the relevant employees or teams is done manually.
With the AI we can tag each email with its respective subject and using flow we can assign the email to the relevant employee or team.
So, I created a very simple excel sheet with examples I thought would cover most possibilities:
Since it was so accurate, I received a very high-performance rate-
This is what it looks like when I tested it, with a sample email-
I used again the word ‘position’, this time in a totally different meaning, and the AI tool identified it to be a support email and not a HR email-
This means that once we receive an email, the AI model will tag it with ‘job’ or ‘support’ tags and the Flow can assign it to the HR or Support teams, respectively.
More ideas for other AI model types in my next article.