Title

Background / Big Picture:

Wrench.ai offers a feature that allows users to drop files and documents into vector storage so that AI agents can have knowledge from outside its model, such as proprietary product information or documents provided by users. Wrench.ai automatically parses and chunks your documents, creates and stores the embeddings, and use both vector and keyword search to retrieve relevant content to answer user queries.

General Guidance

Talking to AI agents is similar to talking to humans. If you provide clear context and explain what you want to achieve, your human friend will understand you better, leading to a healthier conversation. Similarly, AI agents will attempt to make the best guess to determine your intention and the context needed to answer your questions. Therefore, it's always beneficial to provide well-formatted data when interacting with AI agents.

How You Want to Format Your Data

(Image A - Random Image from Google)

(Image B - Random Image from Google)

Our file search feature works significantly better with image A compared to image B.

Factors that Improve File Search Performance:

• Empty Cell Values: If many cell values are empty, it can confuse the AI when interpreting the data.

• Column Headers: It's best to have the first row serve as a column header.

• Use CSV over Excel: Using CSV can relieve some headaches automatically. Extracting content from CSV files works much more smoothly than extracting from Excel files.

• Use Text and Markdown Files over PDF: Extracting content from PDF files often results in poorly formatted text, with some content potentially not being extracted at all.