Proposal: Optional AI Channel Creator with per-channel instructions

[hamidelgendy]

Hi everyone

I’d like to propose an optional AI-assisted way to create and manage Tunarr channels. The goal is not to add AI for its own sake, but to make Tunarr’s scheduling and smart-collection features easier to use.

Main idea

A user could create a channel by describing what they want in natural language:

Create a channel for classic comedy and science-fiction. I prefer shows and movies from the 1980s and 1990s. Play family-friendly content in the morning, sitcoms in the afternoon, and action or science-fiction movies after 8 PM. Avoid repeating the same episode within two weeks and include seasonal programming when appropriate.

Tunarr would use this description to create a draft channel, including:

• Smart collections and content filters
• Daily and weekly programming slots
• Ordering and repeat rules
• Seasonal schedule variations
• A channel name and description
• An optional logo or watermark concept
• A preview of the resulting guide

The generated channel would still use Tunarr’s existing scheduling system. The AI would propose valid Tunarr configuration rather than becoming responsible for playback itself.

Persistent channel instructions

Each channel could have its own editable description containing the user’s preferences and intentions.

For example:

This is a relaxing evening channel. Prefer mysteries, classic dramas, and documentaries. Avoid children’s content. Show longer movies on Friday and Saturday nights.

The assistant could use these instructions whenever the user asks it to update or rebuild the channel. Users could also choose whether the channel is updated manually or automatically at specific times.

Before applying changes, Tunarr should show what will be added, removed, or rescheduled. Automatic changes should be optional and recoverable through backups or rollback.

Event-aware programming

With explicit permission, the assistant could search online for relevant events and use them when planning the schedule.

The user could optionally provide a country, city, or timezone. The assistant might recognize:

• Local and national holidays
• Religious or cultural occasions
• Film and television anniversaries
• Major sports or entertainment events
• Seasonal events such as Halloween, Ramadan, Christmas, or summer holidays

It would then match those events against content already available in the user’s media library.

For example, it could suggest space-related movies during an important spaceflight anniversary, football documentaries around a major tournament, or suitable family programming during a local holiday.

Online access, location information, and automatic event-based changes should all be clearly optional. The assistant should not download media or add unavailable content.

Suggested implementation phases

A reasonable first version could:

1. Accept a natural-language channel description.
2. Generate a complete draft using existing smart collections and scheduling tools.
3. Validate the result and show a guide preview.
4. Require the user to approve it before creating the channel.

Later versions could add persistent per-channel instructions, scheduled refreshes, event-aware recommendations, and optional branding generation.

The feature should be disabled by default, support user-provided AI credentials or local models where practical, explain what metadata is shared, and never upload media files.

Would this concept fit Tunarr’s direction? If so, what would maintainers consider the best initial scope: a design document, a complete draft-channel creator, or a smaller smart-collection and schedule assistant?