Skip to main content

Parallel

Shanone’s Parallel integration gives your agent 7 tools for Parallel’s web search, structured find-and-extract, chat, and asynchronous research/monitoring APIs — built for agentic workloads that need grounded, real-time web data.

Getting Started

1

Create a Parallel API key

Generate a key from the Parallel platform.
2

Add the key in Shanone

Go to Integrations in the Shanone dashboard, select Parallel, and paste in your API key.
3

Retry your request

Once saved, shanone_execute_tool calls for parallel_* tools will succeed.

Available Tools

Shanone provides 7 tools for Parallel, organized into these categories:
parallel_search, parallel_extract — run a web search optimized for LLM consumption, or extract structured content from specific URLs
parallel_findall — discover entities matching a description across the web (e.g. “find companies that do X”), returning a structured list
parallel_task, parallel_chat — run a long-running research task with citations, or have a grounded conversation backed by live web search
parallel_monitor_create, parallel_monitor_list, parallel_monitor_events, parallel_monitor_delete — set up a standing web monitor for a topic and poll for new events

Common Use Cases

Grounded research reports

Run parallel_task on a complex question and get a cited, structured answer back

Lead / entity discovery

Use parallel_findall to build a list of companies or people matching specific criteria

Competitive monitoring

Create a monitor for a competitor or topic and check parallel_monitor_events on a schedule

Structured data extraction

Pull specific fields out of a set of URLs with parallel_extract instead of scraping manually

Troubleshooting

Research tasks run asynchronously and can take longer than a single request for complex queries — check the tool’s response for a task/run ID and poll rather than expecting an immediate synchronous result.
Monitors need time to run their first scan cycle. Wait a few minutes after parallel_monitor_create before expecting events, then poll periodically.
Extraction quality depends on the page being accessible and reasonably structured; for JS-heavy or paywalled pages, results may be incomplete even with a valid API key.