As of 13 August 2026, AI can add subtitles to a client video.

What to actually do

1. Hand it to a person

   The route this page recommends

   A person who owns the outcome does this end to end, worth it when the failure is dear.

2. Use a tool built for this

   Second choice

3. Do it yourself

   The distant third

   A chat interface, chat-fluent skill, and roughly 30 minutes until you can act on the result.

   How to actually do it

   1. Open the client video, confirm the agreed delivery format and gather the client's spelling list, preferred caption style and any instructions about speaker labels.
   2. Confirm that the client has authorised you to upload or process the footage with an AI service, and remove unrelated private footage if it is not needed.
   3. Upload the video to a chatbot that accepts the file, or create a rough transcript first and paste the transcript with the audio timings into the prompt.
   4. Paste the prompt, fill in the format, language, caption limits and glossary, then save the returned SRT or VTT file without changing the uncertain words list.
   5. Open the video and subtitle file in your video editor or subtitle editor, check that the file imports correctly, and apply the client's font, position, colour and safe-area requirements.
   6. Watch the complete subtitled video with sound on, compare every caption with the audio, correct names, punctuation, line breaks and timing, then export the requested caption file or burn the captions into the final video.
   7. Send the finished file and, where useful, the separate subtitle file to the client after checking that the exported video still has readable captions on the intended playback screen.

   Prompt

   I am adding subtitles to a client video for freelance work. Use the attached video or transcript to produce an accurate subtitle draft in [SRT or VTT] format.
   
   Requirements:
   - Language: [language and regional spelling]
   - Maximum characters per line: [number, or leave blank if unknown]
   - Maximum lines on screen: [number, or leave blank if unknown]
   - Client style requirements: [paste them here]
   - Names, brands, technical terms and spellings to preserve: [paste glossary here]
   - Speaker labels required: [yes or no]
   
   Transcribe only what is audible. Do not invent, silently correct, summarise or rewrite the spoken words. Mark unclear speech as [inaudible] rather than guessing. Keep subtitle breaks natural, do not split names or short phrases unnecessarily, and keep each subtitle aligned to the speech. Return the subtitle file in a code block, followed by a separate list of any uncertain words, names or timings. Do not claim that the subtitles are final until they have been checked against the video.

   Open it prefilled in ChatGPT or Claude, or copy it into Gemini, which takes no prefill link.