Agentic Search API Playground
The Agentic Search API playground is a no-code interface built into the Coresignal dashboard that lets anyone explore the Agentic Search API.
Whether you're a developer evaluating a new integration, a recruiter testing search criteria, or an analyst exploring what's possible, the playground gives you immediate, hands-on access to Coresignal's professional data. Describe what you are looking for in plain language, configure a few parameters, and see real results in seconds.
How it works
Select an endpoint
Choose Fast or Reasoning endpoint based on your use case.
Fast (/v2/agentic_search/fast) is optimized for speed and cost. With a simplified schema, it quickly and efficiently translates your prompt into a query, making it well suited for high-volume, programmatic workloads such as AI agents, automated data pipelines, and product search features. It supports up to 1 request per second.
Reasoning (/v2/agentic_search/reasoning) is optimized for accuracy on complex queries. It uses the full data schema, automatically infers the entity type from your prompt, and supports an optional clarification flow that lets the engine ask follow-up questions before generating a query for more precise results. It is best suited for exploratory searches, complex multi-agent setups, and use cases where precision matters more than speed. Subject to a rate limit of 10 requests per hour.
Configure parameters
Adjust the parameters to control the results returned. See the parameters table below.
Once ready, click Submit request.
Clarify the search
Clarification is available on Reasoning. If your prompt is complex or ambiguous, the engine may pause before executing the search and ask one or more follow-up questions. Respond with additional context directly in the playground at a second step, and the engine will use your answers to refine the query before returning results. Clarification steps do not consume credits.
Review the response
The playground shows:
Results count – The total number of matching records found in the database.
Live request preview – A copy of the outgoing API request in your choice of cURL, Python, or Node.js, ready to use in your own integration.
Response output – Either the generated Elasticsearch DSL query or a preview of the matching data records, depending on your selected return mode.
Parameters
Control the response results using the available parameters below:
Entity
Choose the data type you are interested in: company, employee, or job. In a Reasoning request, the entity is automatically inferred from the context of your prompt.
Threshold
Set the semantic similarity threshold. Lower values return broader results with more synonyms included, while higher values return more closely matching results.
Return data
Set the response mode. When toggled off, the response includes the generated Elasticsearch DSL query, which can be used for sending multi-source queries. When toggled on, the response includes matching data records directly.
Number of results (Return data: on)
Set the maximum number of results to return in a response, with a limit of up to 100 results.
Clarification (Reasoning only)
When enabled, the model may ask follow-up questions if the prompt is complex or ambiguous, helping it generate more accurate results before executing the search.
Sample prompts
The playground includes ready-to-use sample prompts to help you explore the API without needing to write your own from scratch. Here are a few examples:
Find Senior Product Managers at fintech companies in London with 5+ years of experience.
Find B2B software companies headquartered in Germany with 50-500 employees, founded after 2015.
Find active job postings for Data Engineer roles at financial service companies in London or Amsterdam posted in the last 30 days.
You can modify any sample prompt to refine your search.
Credits
Each submitted request in the playground consumes search credits, following the same credit logic as the Agentic Search API. Credit cost depends on the endpoint used and the number of results returned. You will see the cost of your request in the prompt field.
Clarification steps do not consume credits.
Elasticsearch query (Return data: off)
2
10
1–20 results
2
10
21–40 results
4
12
41–60 results
6
14
61–80 results
8
16
81–100 results
10
18
Last updated
Was this helpful?