Skip to main content
The Tess chat lets you attach files so the AI can read, analyze, summarize, extract information, transcribe content, or help you work with data and code. You can upload documents, spreadsheets, presentations, media files, images, and many programming formats. The type of file you upload defines the processing workflow used by the platform.

​
What is it?

File support in chat lets you include external content directly in the conversation as context for the AI. In practice, you can:
  • Summarize a PDF or Word document;
  • Analyze data from a CSV or Excel spreadsheet;
  • Extract information from a presentation;
  • Transcribe and analyze audio and video;
  • Review, explain, or fix code files;
  • Ask questions about images attached to the chat.
After attaching the file, send a Prompt explaining what you want to do. Tess will use the available content within the context window and the workflow that matches the format to generate the response.

​
How to use it?

  1. Attach a file to the chat.
  2. Wait for processing to finish before sending more complex requests.
  3. Write a clear Prompt indicating the expected result.
  4. If needed, ask follow-up questions in the same conversation to go deeper into the file analysis.

​
Supported formats

​
Documents and data

These files can be used for tasks such as summarization, information extraction, content comparison, data analysis, and creating responses based on the uploaded material.

​
Audio and video

Audio and video files are processed with transcription via Deepgram. After processing, you can ask questions about the spoken content, request summaries, identify topics, and extract decisions or next steps.
Media processing may consume more credits than text files, because it involves transcription before AI analysis.

​
Images — available in chat

In chat, you can attach images and ask the AI to describe them, extract visual information, compare elements, or generate analyses from the displayed content.

​
Code files

Tess also accepts several code and configuration file formats. All of them follow the same processing workflow for code. With these files, you can ask for explanations, logic review, error identification, documentation, improvement suggestions, and conversion between languages or formats.

​
Deeper explanation

Each format accepted by Tess is associated with a processing workflow. That workflow defines how the content will be prepared for use in chat. In general:
  • Documents, spreadsheets, and code may use smart_search mode, which locates relevant parts of the file before generating the response.
  • Audio and video use transcription via Deepgram before analysis.
  • Images are processed as visual content in chat and in the API.
  • The Web Scrapper feature is available only in the chat context and does not correspond to a file uploaded by the user.

​
Practical examples

​
Summarize a contract in PDF

Attach the .pdf file and use:

​
Analyze a results spreadsheet

Attach a .csv or .xlsx file and use:

​
Review a code file

Attach a .py, .js, or another supported format and use:

​
Extract decisions from a recorded meeting

Attach an audio or video file and use:

​
Analyze an image

Attach an image in chat and use:
SMART PROCESSINGWe use smart processing that identifies the same content already processed in Tess to avoid processing it again. So even if the file name is different and the content is the same, once it has been processed once, we will reference the original one.

​
Important notes

  • Uploading and processing files may consume credits, especially in tasks that involve advanced models, analysis of large files, audio, and video.
  • Very large files or files with a large volume of content may take longer to process and may be analyzed in parts, according to the model’s context window.
  • For audio and video, transcription quality may vary depending on language, noise, multiple speakers, and recording quality.
  • Web Scrapper is a chat-specific feature for web content and is not a file format for upload.