Imagine being able to understand what was said in foreign language videos or audios and having your own personal translator that helps you discover what was said through speech recognition. With the help of Tess AI, this is possible.
In this tutorial, we will explore one of the User Inputs in detail. This content is classified as intermediate, so it is essential that you have completed at least the Text Template Tutorial. If you haven’t done so yet, you can access it by clicking here.
If you have already read it, we estimate that you already possess the basic knowledge about creating templates, so we can delve a little deeper into the topic of input groups.
Know How to Use the File Upload Field in Your Templates
To start your discovery in Tess AI, we will address each of the options below and their applications in the "Upload File" field in this tutorial.
- Short Text Field: By selecting this type, you allow the user to type short texts for more objective answers.
- Long Text Field: It is possible to enable the user to insert large blocks of text if your template requires a large volume of information.
- Dropdown: This allows you to create a single selection box where the user can choose one of the options.
- Multiple Choice: This allows you to create a multiple selection box where the user can select more than one option.
- File Upload: This is where you will be able to upload your attached materials.
At this moment, we will select the Upload File field, as shown in the image below:
[IMG]
The field will be intended for situations where you want to send audio or video files, documents, etc., making this a crucial functionality in everyday life, which integrates perfectly with an AI Step, such as reading documents, for example.
To recap the information that will help you organize and bring your template ideas to life, always keep the following steps in mind:
- Develop the idea;
- Create a prompt;
- Test the prompt in the Conversation Copilot;
- Refine the prompt;
- Create the template.
In this tutorial, we will continue from Step 5. If you need to study the previous steps, you can follow the step-by-step guide by clicking here.
1. Creating the Agent
To start with the initial filling: we will begin by defining the type of AI (chat, image, text). The process is simple, and all you need to do is choose from the options. If you want, you can switch between AI models, filtering for a specific model. By default, the "All LLM" option will already be selected.
After that, we will move to the more technical part of the creation process.
2. Prompt
Idea: A template that can analyze an audio or video file in a foreign language and translate it into Portuguese. Just upload the file and receive all the formatted content.
Approved Prompt: “Transcribe every word that is in the **converted-file**.
Note: Do not invent or add words; show exactly what is in the audio. Then, translate everything into Portuguese [pt-BR]. Result:”
3. User Inputs
Notice that we specified only one AI Step, which will be linked with the Input that we will create below. Select the "Upload File" option and fill it out as usual. In this case, the input field will be simply called "File."
**converted-file** will be the product of an analysis of the uploaded file and will need this input, as we will see next.
[IMG]
This process differs from the other input groups because it involves the inclusion of an Advanced AI Step. We will go through each step so you can follow along.
4. AI Step
For this template, an AI Step connecting with another AI, one for audio transcription, will be needed so that it can be converted to text.
[IMG]
Thus, in “Advanced AI Steps,” select the type “Audio Transcription,” and give a name for the transcription, in this case, “Converted File.”
[IMG]
Finally, specify that the imported field is the “**File**,” which must have the same nomenclature as the Input created. This way, whatever the user inputs in the **File** entry will be transcribed. Ensure that the step is set up like this:
Note that if you do not use the AI Step between asterisks in the Prompt, nothing the user uploads will be used by the AI. Therefore, make sure it is filled out exactly as created, including the **asterisks**, or click on “+” to auto-fill.
[IMG]
5. Save
When your agent is ready, simply click “Save” and “Preview” if you want to test it. You can also find it in AI Studio as a draft.
The template of this tutorial was tested with a video in a Chinese dialect and turned out like this:
My Agents
You can validate your prompt after approval by testing, editing, and refining it until you give the model your personal touch.
If you want to publish it, just change the visibility in AI Studio. In this case, the agent is sent for approval.
This is another tutorial for creating templates through Tess AI. Be sure to keep up with our updates and resources as we continue to expand the possibilities of AI.
A world of possibilities awaits you - start creating your AI right now!