Norn API

Norn is the model behind the ideas Native suggests to its customers, more than 2,200 businesses. It does not write posts. It decides what a post should be: the point to make, the angle to take, the format, and why it fits now.

The API opens that up to any product. Send Norn a brand, and it sends back ideas in order, best first. Each idea says why it should work now and cites where it came from, such as a page on the brand’s website or a customer review. The model you already use writes the post from there.

Language models are very good at writing. They are not good at knowing what to talk about. They are trained to predict the most likely next word, so their ideas drift toward the average, and the average is what people scroll past.

Norn is built around a different question: which idea will stand out for this brand, with this audience, at this moment. It answers from what people actually do with ideas: which ones they approve, rewrite or turn down, and how the posts perform once they are out. We call that artificial intuition.

1. A brand. Its website, language and country, and any rules you set. Norn reads the website once and keeps what it learns. See Brands.
2. A request for ideas. How many you want, in which formats, and anything Norn should focus on. See Ideas.
3. Ranked ideas back. Each has a hook, an angle, a reason it fits now, its sources, an outline and a prompt for a generative model.
4. Feedback from you. What people did with each idea and how the post performed. Norn uses it to learn the brand’s taste. See Feedback.

Every idea Native suggests leaves a trail: the idea, what the customer did with it, and how the post performed. That trail is what Norn learns from. When customers turn ideas down and say why, Norn turns the reasons into lessons for that brand. When a post does better than the brand’s usual, Norn notes what made it work.

The API is built to feed the same loop. The more you tell Norn about what happened to its ideas, the better the next ones fit.

• Brands, ideas and feedback make up the first preview. They are what Native runs on today, opened up as an API.
• Predictions for drafts you already have are in development. The Predictions page shows the plan.
• There is no SDK yet. The API is JSON over HTTPS, and every example comes in cURL, TypeScript and Python.
• Pricing is not set yet. You will see the price before you are charged anything.

• QuickstartFrom your credentials to your first ranked ideas.
• IdeasWhat an idea holds, and how to ask for good ones.
• BrandsWhat Norn reads from a website, and how to correct it.
• FAQWhat is live, what it costs, and how Norn differs from a chat model.