The AI Research node (AI Research) runs live online research on a query using Perplexity Sonar models and returns a sourced answer. It performs real web research, so it is a high-cost node.
When to use it
Use it when a task needs current, web-grounded facts: market research, competitor checks, or gathering sources before you write. Feed its output into a Summary Generator Node or Write Article Node to turn findings into content.
Configuration
- Content: the research question or topic. Insert upstream data with variable tags.
- Model: a Perplexity Sonar model:
- Sonar: fast and efficient with real-time web search.
- Sonar Pro: better citations and support for complex queries.
- Sonar Reasoning: lightweight reasoning.
- Sonar Reasoning Pro: stronger reasoning for hard questions.
- Sonar Deep Research: exhaustive multi-source research.
- Search recency: limit results to any time, the past 24 hours, week, month, or year.
- Domain filters: restrict research to specific domains.
- Max tokens and temperature: control response length and variability.
Example
Latest trends in [[topic] from trigger-1]
Pick Sonar Pro and get back a sourced answer, referenced downstream as [Input from research-1]. See Data Flow and Variable Handling.
Local, Cloud, and credits
Runs in both Local vs Cloud execution modes. This is the most expensive AI action node: it costs 40 credits per run in Cloud mode because it performs live research. In Local mode it uses your own Perplexity key.