Use cases

Catch what matters on Hacker News

On Hacker News the link is only half the story. The comments are where people test the claim, share what they use instead and say what is missing. Follow the lists you care about, let Owlie keep the stories on your topics, and read each one with what the discussion concluded.

The recipe

One Operation follows every Hacker News list you choose. These are the settings to choose when you create it; you can change any of them later.

  1. Sources

    Paste a list address: the front page (news.ycombinator.com), Newest, Best, Ask HN, Show HN, or a user’s submissions. Choose what to read from each story: the Post (the linked article, or the text of an Ask HN or Show HN post), the Comments, or both. Owlie checks each list every hour.

  2. Filter

    Choose Keywords for the tools, companies and topics you follow, or AI criteria for something keywords can’t pin down, such as “a launch or discussion about developer tools for small teams”. Keywords are checked after the story is read, so a word that only appears in the comments still matches.

  3. Transform

    Choose Analyze with AI for a short written summary of the story and the discussion. Or choose Structured content and add fields, such as the verdict of the comments, the main criticisms and the alternatives people mention.

  4. Deliver

    Stories land in your Inbox with a link to the thread. Turn on email for the ones you don’t want to miss, or add a webhook to send each one to a team channel or your own app.

Example

Illustrative example with a made-up Show HN post and made-up comments. The Source reads the comments; the fields turn the discussion into a short verdict.

  1. Extracted content

    A Show HN story with its comments, as Owlie reads it.

    news.ycombinator.com/show

    Show HN: Tidepool – self-hosted error tracking in one binary
by mkline · 212 points · 96 comments

Comments:

[jdoe]
  Setup took two minutes. Grouping is
  better than what we had. Does it
  support source maps yet?

  [mkline]
    Not yet, it's next on the list.

[rserra]
  We left Sentry over pricing. One
  binary plus SQLite is the right call
  for small teams.

[akim]
  Retention is hardcoded to 30 days,
  which rules it out for us.

  2. Fields

    Each field says what to extract. The same fields apply to every kept story.

    Fields: what to extract

    {
  "summary": {
    "type": "text",
    "description":
      "What was shared, in one line"
  },
  "verdict": {
    "type": "enum",
    "values": ["positive", "mixed",
      "negative"],
    "description":
      "Overall tone of the comments"
  },
  "praise": {
    "type": "list of text",
    "description":
      "What commenters liked"
  },
  "criticisms": {
    "type": "list of text",
    "description":
      "Concerns and missing features"
  },
  "alternatives": {
    "type": "list of text",
    "description":
      "Other tools people mention"
  }
}

  3. Structured Result

    The discussion in the same shape for every story.

    Result

    {
  "summary":
    "Self-hosted error tracking",
  "verdict": "mixed",
  "praise": [
    "Two-minute setup",
    "Good error grouping"
  ],
  "criticisms": [
    "No source maps yet",
    "Fixed 30-day retention"
  ],
  "alternatives": ["Sentry"]
}

Variations

The same recipe with different Sources and fields.

  • Watch launches in your market

    Follow Show HN with AI criteria that describe your market, and read each launch with what commenters praised and questioned. You see new competitors early and learn what people expect from them.

  • Hear when your product comes up

    Follow the front page and Newest with your product and company names as keywords. Because keywords are checked after the story is read, a mention in the comments counts too.

  • Learn from Ask HN

    Follow Ask HN with Comments only and a structured field for the recommendations people give. Questions such as “what do you use for…” turn into lists of tools with the reasons behind them.

  • Read the front page in minutes

    Follow the front page with no filter and a one-paragraph summary of the post and the discussion. Each morning, your Inbox has every story that reached it, with the gist of what people said.

Good to know

What this recipe can and can’t do today.

  • The first 30 stories of each list

    Each hourly check reads the first 30 stories of a list, the size of the front page. That covers the front page, Best, Ask HN and Show HN well. Newest can move faster than that at busy times, so some stories pass between checks.

  • Comments are read after about two hours

    A story read with its comments waits until it is about two hours old, so the discussion has started. For Newest, choose Post only: by the time a story is two hours old it has usually left the newest 30.

  • The top of the discussion, not every reply

    Owlie reads up to 100 comments: the first 30 top-level comments in Hacker News order, and their direct replies. Long threads are summarized from that part of the discussion.

  • Lists, not search

    Owlie follows the lists you add. It doesn’t search all of Hacker News for a word or filter by points, and it doesn’t follow a single thread. Use Best or the front page when you only want popular stories.

  • What it costs

    Keyword checks are free. Each Result costs 1 credit plus its AI usage, so at least 2 credits. Reading the comments makes each story longer, so AI steps may use more credits. AI criteria cost at least 1 credit for every story they read. New accounts start with free credits.

Questions

Short answers to what people ask most.

  • Do I get the whole front page when I add it?

    No. The first check reads the top story, so you see a Result straight away. After that, Owlie reads each story that newly appears in the list.

  • What about a story that reaches the front page later?

    Owlie still catches it. It remembers which stories it has already read instead of when it last checked, so an older story that climbs into the list is read when it appears.

  • What if the linked article can’t be read?

    Owlie reads pages that load without JavaScript or a login. When you read both the Post and the Comments and the article can’t be read, the Result is based on the discussion.

  • Can I follow Hacker News together with other sources?

    Yes. One Operation can follow Hacker News lists, YouTube channels, RSS feeds, subreddits and podcasts together, with one filter and one set of fields.

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