How it Works: The API as a Tool
In modern AI frameworks like LangChain or with OpenAI's function calling, you can define "tools" that an agent can decide to use to answer a question. The Whisper API endpoints are perfect tools for any agent tasked with internet investigation.
An agent's reasoning process might look like this:
- User Prompt: "Investigate the domain
suspicious-login.net
and its infrastructure." - Agent Thought: "My first step is to get general information about this domain. I will use the
get_domain_intelligence
tool." - Agent Action: The agent calls
GET /intelligence/v1/domain/suspicious-login.net
. - Agent Observation: The agent receives the JSON response. It sees the domain resolves to IP
4.5.6.7
and was created yesterday. - Agent Thought: "Now I need to understand the reputation of the IP
4.5.6.7
. I will use theget_ip_intelligence
tool." - Agent Action: The agent calls
GET /intelligence/v1/ip/4.5.6.7
. - Agent Observation: The agent sees the IP has a high
risk_score
and is linked to other suspicious domains. - Final Answer: The agent synthesizes these findings into a human-readable summary, concluding that the domain is high-risk.