The short version
Every puzzle on The Daily Quip is built with the help of AI. A human (that’s our founder, Ken) decides the editorial direction, the voice, the structure, and the rules for what makes a story worth turning into a game. The AI applies those rules and handles the production — generating the puzzle content at scale, every day. Think of it the way you’d think of a newspaper using a printing press: the editorial judgment is human, the machinery is modern.
What’s AI-generated
The puzzle content itself — the clues, the connections, the encoded headlines, the indirect angles, the riddles, the hot takes — is generated by Claude, Anthropic’s AI. So are the daily columns, the Shonen Showdown battles and the After Hours rounds. The daily cartoon and the game illustrations are generated with OpenAI image models.
The news puzzles start from current news stories gathered from third-party news services and feeds, including NewsData.io. Those stories are human journalism; the AI uses them as the basis for each day’s games. The comics games draw on the Grand Comics Database, and the anime games on AniList.
What’s human
The things that make TDQ feel like TDQ:
- The editorial philosophy — what kinds of stories belong here, and why
- The voice and tone of each game format
- The structure of every puzzle type
- The editorial personas (more on those below)
- Approving every Shonen Showdown battle and After Hours round before it runs
- The decision to build a puzzle platform that explains today, not just distracts from it
None of that comes from a prompt. That’s the product.
The voices: Silas, Kai, Aisha, and Mateo
Our editorial staff are personas — defined characters with distinct voices, points of view, and cultural sensibilities. Silas will say the thing nobody else will. Kai keeps it street-level and lateral. Aisha builds the case. Mateo finds the thread that connects everything.
Their perspectives and personalities are human-designed. Their daily writing is AI-generated. Think of them the way you’d think of a columnist: the voice is real, the byline is intentional, the production is industrial. They’re not fake people — they’re editorial identities, in the tradition of every publication that’s ever given a column a character.
The technical details, if you want them
| Text model | Claude API (Anthropic) — claude-sonnet-5-5 |
|---|---|
| OpenAI models | OpenAI — gpt-6-astra (cartoon concepts), gpt-image-2.5-sunburst (cartoon art), gpt-image-2.5-flare (Cover or Clickbait? covers), gpt-image-2 (Shonen Showdown art) |
| News sources | Third-party news services and feeds, including NewsData.io |
| Reference data | Grand Comics Database (comics), AniList (anime) |
| What the AI does | Scores each day’s candidate stories, generates puzzle content from them, and checks it against editorial quality standards before publication |
| What the AI doesn’t do | Set the editorial rules, decide the voice, design the game formats, or make structural decisions about the platform |
Why we’re telling you this
Because you deserve to know. Also because the EU AI Act requires it (Article 50, if you’re the type who looks these things up). But mostly because transparency about how something is made doesn’t diminish it — it explains it. The Daily Quip is a human editorial product that uses AI to publish at daily scale. That’s the honest version, and we think it’s a good one.
Questions? Thoughts? Strong opinions about AI in media? We’re at kcdailyquip@gmail.com.