AI builders and agent developers: One call: is it a real brand, and how hot?
Give an agent grounded facts (maker, products, rank, momentum) instead of letting it guess.
When does this come up?
An agent or app meets a company or product name and needs to know what it is.
What is the mental model?
Give the model grounded facts and a date, and it stops guessing.
- Signal: A name the agent has just met
- Filter: Is it a real brand? Who makes it?
- Decision: Answer with facts, not a guess
- Action: Cite the number and its date
How does this work in DawnTrend?
- Ask by name: The search box and the find endpoint match brands and products, and a product tells you which company makes it.
- Read the brand: One JSON document has the score, the signals, the ranks, the parent and the products.
- Give it to the model: Every page also has a Markdown version and the site publishes an llms.txt, so a crawler or an agent reads the same facts.
What does this cost to run?
A free account has 2,000 API calls and 250 search credits a day. Reads from the index use no search credits.
| Step | Call | Calls | Search credits |
|---|---|---|---|
| Match the name | GET /dawn/find?q=claude | 1 | 0 |
| Read the brand | GET /brand?slug=claude | 1 | 0 |
| Whole workflow | 2 | 0 |
What does this not tell you?
- New names are identified on demand and remembered; a name we cannot verify is not shown as a brand.
- Free accounts have a daily allowance; the API needs a key.
How do I get the same data as JSON?
GET /api/v1/brand?slug=claude. Free accounts get an API key.