Web research API for AI assistants

The web research API takes a question and returns a written answer researched from multiple web sources, with links and citations. Your assistant uses it to pull together a short brief on a topic you need to know about.

What it does

You pass a question as plain text and choose a research effort level, typically lite or standard. The provider searches across multiple sources, reads the results, and writes an answer that cites where each claim came from. The reply includes the answer text and a list of sources with their URLs and titles.

Some providers support optional structured output, so you can request the answer in a specific JSON shape, and optional source or recency controls to limit which sites or time ranges the research draws from.

How a Joshu assistant uses it

When you are about to meet a new client or partner, Joshu researches the company and the people on the invite, then writes a short note with what they do, recent news, and who you already know there.

When you hand Joshu a question that needs more than one source, such as comparing pricing models or checking what changed in a regulation, it researches the question and replies with the answer and the links, so you can scan the sources yourself if you want.

When you want a weekly digest of a topic you follow, Joshu runs the same research query every Monday and emails you the brief with the new sources from that week.

Choosing a provider

Providers differ in how many sources they search, how they rank relevance, and whether they support structured output or source controls. The lite research mode is faster and cheaper, while the standard mode searches more sources and writes a longer answer.

curl https://hello.joshu.me/api/router/call \
  -H "Authorization: Bearer $JOSHU_ROUTER_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "endpoint": "you.web.research",
    "params": {
      "input": "What is the purpose of example.com?",
      "research_effort": "lite"
    }
  }'