Bring the source into the agent's context.
A news search API connects an agent's question to material it can inspect. AriseLabs searches continuously crawled news and market pages. Standard search returns indexed article text with each source URL, title, and available publication time.
Use this material to investigate a company, follow a developing story, or assemble context for a prediction. Keep each passage attached to its source so the final answer can cite the evidence behind a claim.
Start with a focused query.
Name the entities and concepts that matter. The request below asks for five results about Blackwell supply, with a seven-day recency window and no historical cutoff.
curl --fail-with-body https://api.ariselabs-search-api.com/v1/search \
-H "Authorization: Bearer $ARISELABS_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"query": "nvidia blackwell supply",
"mode": "standard",
"k": 5,
"recency": "7d"
}'Set ARISELABS_API_KEY to an API key from your console. Run this on your server or in a local terminal.
Turn the response into useful context.
| Field | Use it for |
|---|---|
data[].text | Evidence the agent can read directly. |
data[].url | Source attribution and citations in the final answer. |
data[].title | A readable label for each source. |
data[].published_at | Publication context when available; it can differ from indexing time. |
search_metadata | Retrieval diagnostics, including the returned count and degradation signals. |
usage.credits | The credits used by the request. |
Deduplicate overlapping stories before adding them to context. Ask the agent to separate reported facts from its own inference, cite the sources it uses, and say when the retrieved evidence is insufficient.
Choose how fresh the evidence needs to be.
With no as_of_date, standard search uses the live index. It reflects what has been crawled and indexed, so it should not be treated as an exhaustive real-time wire feed. A missing story may not yet be in the corpus.
For a known URL's current contents, the Fetch API provides a separate request. Advanced search can supplement retrieval with live web sources. For a historical evaluation, use point-in-time standard search and keep the indexed text returned at that cutoff.
Fit search into your own agent loop.
Call POST /v1/search from a server-side tool with a bearer API key and JSON body. You can use ordinary HTTP clients in Python or JavaScript; an agent framework is not required. Pass the returned evidence into the model or orchestration system you already use.
Check the API reference for supported parameters and limits, and pricing for credit costs. For the retrieval design underneath the API, read how we built our live ANN index.