Use cases

Follow AI and developer news

AI moves faster than anyone can watch. Follow the channels and blogs you trust, let Owlie skip the videos that don’t matter to you, and read the announcements and takeaways from the ones that do.

The recipe

One Operation follows every channel and blog. These are the settings to choose when you create it; you can change any of them later.

  1. Sources

    Add the YouTube channels that cover models, APIs and developer tools, and the RSS feeds of the blogs and changelogs you read. Owlie checks them every hour for new videos and posts.

  2. Filter

    Choose AI criteria and describe what counts, for example “new models, APIs, SDKs or developer tools; skip opinion pieces and tutorials”. The AI reads each new item and keeps only the matches.

  3. Transform

    Choose Analyze with AI and write the instruction: what was announced, what changed for developers, and what to try. Every Result follows it.

  4. Deliver

    Results land in your Inbox under a Label of your choice. Turn on email to hear about each one, or send them to a webhook, for example to post in a team channel.

Example

Illustrative example with a made-up video and made-up products. The AI criteria let the video through; the instruction decides what the Result says.

  1. Extracted content

    The transcript Owlie extracted from a new video on an AI news channel.

    YouTube transcript

    This week Northwind released Lumen 2, an open-weight model aimed at coding. It is about half the size of Lumen 1 and scores higher on their coding benchmarks. The weights are on their model hub, and the API is live today at a lower price per token.

They also shipped version 3 of their Python SDK. Streaming is now the default, and the old complete() method is gone, so you will need to update your code before upgrading…

  2. Instruction

    Written once in the Operation and applied to every video that passes the filter.

    Analyze with AI

    List what was announced, one line each. Then say what changes for a developer who already uses these tools, and suggest one thing worth trying this week. Skip sponsors and channel news. Keep it under 120 words.

  3. Result

    Ready in your Inbox, in an email, or at your webhook.

    Result

    Announced
- Lumen 2: open-weight coding model, half the size of Lumen 1
- Lumen 2 API, cheaper per token
- Python SDK v3, streaming by default

What changes for you
SDK v3 removes complete(). Pin v2 or move to the streaming calls before you upgrade.

Try this week
Run Lumen 2 on a few of your own coding prompts and compare the cost.

Variations

The same recipe with different Sources and fields.

  • Release notes for your stack

    Add the release feeds of the frameworks and libraries you depend on, and ask for breaking changes and upgrade steps only. Use Structured content instead to get the same fields for every release.

  • A research digest

    Follow channels that explain papers and the blogs of research labs. Ask for the claim, the method in one sentence and whether code or weights are available.

  • Talks from conferences

    Follow a conference’s YouTube channel during its event week and filter for the topics you work on, so you watch the two talks that matter instead of forty.

  • A team briefing

    Send every Result to a webhook that posts in your team’s chat, and give the Operation a Label so the briefing is easy to find in your Inbox too.

Good to know

What this recipe can and can’t do today.

  • Owlie reads the transcript

    For a YouTube video, the AI reads the transcript, in English when one is available. Slides, code on screen and demos that are never said out loud are not part of it.

  • Live streams and Shorts are skipped

    Owlie follows a channel’s regular uploads. Live streams and Shorts can’t be read.

  • New items are checked every hour

    Owlie looks for new videos and posts at the top of every hour, so a Result can arrive up to about an hour after something is published, plus the time it takes to process.

  • What it costs

    The AI criteria cost at least 1 credit for every new item they read, kept or not. Each Result then costs 1 credit plus its AI usage, so at least 2 credits. New accounts start with free credits.

Questions

Short answers to what people ask most.

  • Can I summarize YouTube videos automatically?

    Yes. Add a channel and Owlie turns every new upload into a Result that follows your instruction. You can also add a single video to get one Result now.

  • Can I mix YouTube channels and blogs in one Operation?

    Yes. One Operation can follow YouTube channels, RSS feeds, podcasts and subreddits at the same time, with one filter and one instruction for all of them.

  • Do I need AI criteria, or are keywords enough?

    Keyword checks are free and match the words you list, such as a product name, in the item’s text or transcript. AI criteria understand meaning, such as “announcements, not opinions”, and cost a little for every item they read.

  • Can I see why a video was kept or skipped?

    Yes. Each Operation has an Activity tab that lists every check, the new items it found and what happened to each one.

More use cases

Other ways to put Owlie to work.

Set this up in a few minutes.

Create an account, add a source, and choose the fields Owlie extracts from every new item.

Create accountSign in