Step 01
Prepare your source material
Use clear, well-formatted files where possible. Trad AI supports DOCX, PPTX, PDF, XLSX, and plain text input. Remove any artefacts or embedded images that could interfere with extraction.
How it works
Text guide
A step-by-step walkthrough of preparing your content, configuring the workspace, and exporting high-quality translations with Trad AI.
Step 02
Configure language settings
Choose the source and target languages, then optionally add tone or domain hints. These settings help the model tune terminology and tone for your project.
Step 03
Translate
Click Translate to generate the draft. Trad AI uses the OpenAI model linked to your API key, so usage and quality align with your own account configuration.
Step 04
Export & reuse
Download the translated document and the TMX translation memory. Import the TMX into your CAT tools to keep terminology aligned and accelerate future projects.
How token limits work
Trad AI uses your OpenAI API key, so translation limits follow the model you select. Each model processes a maximum number of tokens per request, covering both source and translated content.
- 1 token is roughly equal to 3–4 characters of English text.
- The limit includes both the text you send and the output you receive.
- If you exceed the limit, OpenAI returns an error such as “context length exceeded”.
Split large documents into smaller sections or switch to a model with a higher token allowance to continue.
Quick checklist
- Confirm that source and target languages are different.
- Include glossary hints via the Tone and Domain fields where useful.
- Download the TMX file after each project to build your translation memory.
Keep learning
Jump to the FAQ for detailed answers, watch the video tutorial, or contact our team for tailored advice. You can read the full version of the How it works guide here.