Data & AI

What AfterStream processes

"AfterStream processes this" and "AfterStream sends this to the AI provider" are not the same thing. To analyze a completed Twitch or Kick stream, AfterStream processes:

• Stream information you submit — the VOD, platform, title, streamer name, and duration.
• Chat activity from the stream — individual chat messages and aggregate chat statistics calculated from them.
• Engagement signals calculated from that chat activity, used to identify moments where activity rose noticeably above the stream’s baseline.
• Clips and thumbnails, which AfterStream creates from the stream’s video using its own processing pipeline.
• Generated content — the AI summary, explanations, and publishing suggestions described below.

For the full list of what AfterStream stores and why, see Privacy.

What goes to the AI provider — and what doesn’t

AfterStream currently uses OpenAI to generate stream summaries, highlight explanations, and content suggestions. Only a selected, high-level slice of what AfterStream processes is sent to OpenAI for this.

How AI helps

AfterStream first identifies candidate highlight moments using audience and chat activity signals — not by having an AI model watch the full stream. Once those moments are identified and their clips and thumbnails are prepared, AfterStream sends the selected stream and engagement information above to OpenAI to help interpret those moments and prepare a written report and publishing suggestions.

1. Stream is submitted and processed
2. Chat activity is analyzed into engagement signals
3. Moments with the strongest activity become highlight windows
4. Clips and thumbnails are prepared for those windows
5. Selected signals are sent to OpenAI for interpretation
6. AI prepares a summary, explanations, and publishing suggestions
7. You review and decide what to use

From this, AI currently generates:

• A stream summary, most exciting moment, funniest moment, and growth recommendations.
• For each highlight: an explanation, why it matters, a YouTube title suggestion, and a TikTok hook and caption.

Creator control

AfterStream recommends. You decide. AI output is preparatory — it may need review or editing, and AfterStream does not automatically post, schedule, or publish anything on your behalf.

• PreviewWatch a highlight’s clip and read its generated content before using it.
• EditChange generated titles, captions, and explanations to match your voice.
• CopyCopy publishing suggestions for use wherever you post.
• DownloadDownload the clips AfterStream creates.
• DeleteDelete a highlight, a stream, or your entire account.

Retention and model training

AfterStream does not train its own AI models using your stream data. OpenAI also does not use inputs or outputs submitted through its API to train or improve its models by default, unless the API customer explicitly chooses to share that data with OpenAI.

Under OpenAI's standard API data controls, information from API requests may be retained in abuse-monitoring logs for up to 30 days, unless OpenAI is legally required to retain it for longer. OpenAI also offers additional data-retention controls to eligible API customers.

Specific retention periods within AfterStream for stream, chat, clip, thumbnail, and AI-report data are still being finalized. Deleting a highlight, a stream, or your account works as described on Privacy — that page also has the current detail on storage, deletion, and service providers.