How to configure automations with AI in TimeTonic

Setting Up AI Automations in Timetonic.
Creating an automation with AI is designed to optimize our workspaces and allows us to delegate some of our tasks, such as generating response suggestions, summary reports, data field entries, data reformatting like phone numbers, or translations into another language. The list is long and inspiration comes with use, aiming to meet your own needs.

To design it, it consists of four essential phases.
We need to determine the trigger. Next, configure the role you will assign to the AI. Select the data from your workspace that the AI will process. And finally, insert a new column where the AI’s response will be integrated. Then, you will validate the automation and see the result brought by the AI.
At that point, you will either be satisfied or need to iterate several times by modifying the prompt given to the AI to get the desired result.

Let’s get started concretely.

In our workspace, let's configure our AI automation.
In the horizontal menu bar, click on Automation.
Click on Automation with Timetonic plus AI.
Add a new automation. Name your automation.
Select a trigger.

In the example, I will use a copy-paste of a video transcription for subtitle generation to create a tutorial title.

I won’t add additional conditions, but I could.
I will select the automation by ChatGPT.
I will enter the OpenAI API key, which is available in your OpenAI account under the User tab, API Key.
You can generate one, or if you already have one
available, use it.

Here, you can enter the organization ID.
It is available in your OpenAI account under settings.
It is an optional identifier.
In this case, I will not fill it in.
Here, I select the latest available model of ChatGPT.
I will enter the role I will give to the AI.
I give it a role. It is video creation assistant.
It will receive a video subtitle transcription.
With the elements of this information, I ask it to generate
the title of the video for me.
This video will be published on YouTube.
I will indicate where the question comes from.
I use the field value search
of the transcription.
You can indicate a maximum number of tokens to be used for generating the AI response here.
In this case, I will not do it.
As an indication, know that
four characters correspond to one token,
which allows you to manage the limitation

and the subscription plan in OpenAI.
Then you will select the Create column
to paste the response and you validate.
For the sake of detail, I like to specify in my field
that the column was generated by AI.

Let’s do the test now.
So, let’s paste the subtitle description of this video obtained
into the AI.
Give it some time.
And there, my title is generated automatically.
If it suits me, that’s great.
If it does not suit me,
I will readjust the description of the role I give to the AI and iterate
once again to get
as close as possible to the result I expect.
From the same question given to the AI, I could duplicate my
AI automation to then generate the description
and any other complementary needs
of activity, to my creation process.

To complete the information, it will be noted that when creating
a new automation,
we can also choose
an AI automation by Mistral AI.
The configuration is identical, the model is different.
You can find the detailed information in the documentation
regarding the three models.
This is the only configuration difference.

As you will understand, AI automations
will allow you to enhance your processes
by targeting the interesting steps to delegate to the AI.
You build decision support tools,
personalized accompaniment,
and reduce the entries that can be entrusted to it in our Timetonic tool.

‍