The Consulate faced a tight timeline. With participating Argentine companies confirmed only weeks in advance, there was little room for slow, manual outreach. Traditional outbound prospecting—researching companies one by one, finding the right contacts, and qualifying them—was becoming a bottleneck.
Within one month, the Consulate needed to define the ideal attendee profile, identify a large volume of qualified U.S. businesses across Georgia, find the right decision-maker in each company, and send invitations fast enough to ensure a strong, relevant audience.
At a glance, the core orchestration logic is implemented in Python, ensuring that every component is called and executed at the precise time needed within the prospecting workflow.
The agent operates using the following key technical elements.
Orchestration and Logic: Python manages the sequential and conditional execution of tasks.
Input Handling: The agent takes its initial selection criteria and instructions as input, typically provided via the argument line as well as configured prompt.
Core Intelligence: A Large Language Model (LLM) is equipped with specific tools to perform research and data management:
Web Search: Used for research during the Discover and Enrich stages to identify companies and gather additional details.
Google Spreadsheets: Employed to record the results, facilitating the automatic generation of the detailed prospect list shared with the Consulate team.
Deployment: The agent is deployed on the AWS cloud using a serverless architecture, which provides scalability and cost-efficiency.
Parallelism: The architecture allows the agent to be configured for multiple, simultaneous executions with different initial instructions, favoring parallelism to accelerate the prospecting process.




