Skip to main content

​
AI Research

AI Research is the product area for end-to-end research workflows. Instead of calling a narrow lookup endpoint, you provide a task and a settings object. NeuralVerge then runs a multi-step process that can search for information, evaluate sources, synthesize findings, and generate a final response.

​
What AI Research is for

Use AI Research when the task requires:
  • multiple searches
  • reasoning across multiple sources
  • a concise analyst-style summary
  • structured outputs for downstream systems
  • a reusable research pipeline instead of a single lookup
Common examples:
  • investigate a company and identify operational risks
  • compile a market overview from public sources
  • summarize external sentiment and reputation signals
  • verify claims about a person or organization
  • produce structured intelligence for internal tooling

​
How it works

A typical research run looks like this:
  1. You send instructions and settings
  2. NeuralVerge starts a session
  3. The workflow searches and analyzes relevant sources
  4. The system synthesizes the result
  5. You retrieve the result from the session status endpoint

​
Inputs

The main AI Research endpoint accepts:
  • instructions for the task itself
  • settings for workflow behavior and model configuration
Depending on your integration, settings can control things like:
  • country or localization behavior
  • whether search is enabled
  • model selection
  • extraction schema usage

​
Outputs

Research results can include:
  • a human summary for UI display
  • a machine object for structured consumption
  • total points or usage metadata
  • session state and progress

​
When to choose AI Research

Choose AI Research when:
  • the task is exploratory
  • you need synthesis, not just retrieval
  • the answer depends on multiple sources
  • you want a richer final report
Do not use AI Research if you already know the exact page to extract from or if you only need a focused provider lookup. In those cases, use AI Extract or Data Sources. In the API Reference, AI Research includes endpoints for:
  • starting a research run
  • lightweight search execution
  • checking session status

​
Typical product pattern

A common integration pattern is:
  • start the run from your backend
  • persist the session_id
  • poll until completion
  • render human
  • store machine for automation or analytics