Case Study: Exploring the US Secret Service with Google Deep Research

Introduction

My Google One subscription includes Deep Research, which, according to its provider,

 … uses AI to conduct research for you, providing a comprehensive report with key findings.

Enterprise and solutions architects must often research unfamiliar topics quickly. Therefore, in order to gauge its value for architects, I asked Google Deep Research to synthesize and expand upon two topics concerning the US Secret Service about which I have written recently: how it could conceivably apply standards-based EA in the aftermath of a near-miss assassination attempt on a presidential candidate; and its use of surveillance drones.

Usage

I used Google Research through the Gemini chatbot as follows.

  1. I asked Gemini to research the US Secret Service
  2. It responded with a research outline, and asked me if I wanted any changes.
  3. I asked for changes, Gemini revised the outline, and I repeated this process until I was satisfied.
  4. I asked Gemini to generate the report.
  5. I asked for changes in the report, Gemini regenerated it, and I repeated this process until I was satisfied.
  6. I clicked on the sharing icon below the onscreen report to export it to a Google Doc.

Here are some excerpts from my dialogue:

Result

Here is an outline of the final version of the report with both section headers and some representative samples of the body text:

  • The United States Secret Service: A Legacy of Service and a Call for Reform

The United States Secret Service (USSS) is a federal law enforcement agency tasked with the crucial dual mission of protecting national leaders and safeguarding the nation’s financial infrastructure. Established in 1865 as a branch of the Department of the Treasury, the agency’s initial focus was to combat the rampant counterfeiting that threatened the nation’s economic stability following the Civil War.¹ This emphasis on economic security as a cornerstone of national security laid the foundation for the agency’s evolution and its present-day role in combating cybercrime and protecting financial systems.²

  • A Historical Perspective: From Counterfeiting to Presidential Protection
  • Roles and Responsibilities: Safeguarding Leaders and Financial Systems
  • Secret Service Budget Breakdown

The Secret Service’s annual budget is a significant investment in national security, reflecting the agency’s wide-ranging responsibilities. In the most recent fiscal year, the budget was approximately $3 billion. ¹¹ A substantial portion of this budget, about 87%, or $2.7 billion, is allocated to operations and support, which includes $1.2 billion for Protective Operations—the division responsible for protecting the President, Vice President, and their families. The remaining $400 million is directed towards procurement, information technology, construction, and research and development. ¹¹

Within the operations and support allocation, the budget is further divided among key protective functions:

  • Protection of Persons and Facilities: This category, with a budget of $907 million, focuses on protecting presidents, vice presidents, and their families. ¹¹
    • Protective Countermeasures: With a budget of $82.5 million, this area concentrates on protecting the President and Vice President at the White House and the Vice President’s residence from emerging threats, including explosive, chemical, biological, radiological, and cyber threats. ¹¹
    • Protective Intelligence: This division, with a budget of $94.6 million, investigates individuals or groups that pose potential threats to the President and other protectees. ¹¹
  • Joining the Ranks: Becoming a Secret Service Agent
  • In the News: Recent Events and Developments
  • A Critical Examination: The Attempted Assassination and Security Failures in Butler, Pennsylvania
    • Key Findings: A Breakdown of Security Lapses

The interim report identifies several critical security failures that demand immediate attention:

Inadequate Planning and Coordination: The Secret Service failed to adequately plan and coordinate security measures with state and local law enforcement agencies. ¹⁹, ²¹ This included a lack of joint meetings and unclear communication regarding assignments and responsibilities, hindering effective collaboration and potentially contributing to confusion and delays in response…

  • Analysis: Systemic Issues and the Need for Reform
    • Further Details from the Report
  • Leveraging Technology for Enhanced Protection: The Role of Surveillance Drones
  • Conclusion: A Legacy of Service, A Call for Reform

…The interim report’s findings provide a roadmap for reform, highlighting the need for enhanced planning and coordination, comprehensive risk assessment, improved sniper deployment, the establishment of a unified command post, and streamlined communication channels. By addressing these issues and embracing new technologies like surveillance drones, the Secret Service can reinforce its legacy of service and ensure the continued safety and security of national leaders and the integrity of the nation’s financial systems.

Analysis

My interaction with the Google Deep Research service through the Gemini chatbot was straightforward. I provided simple instructions and got clear and straightforward responses. The report produced was easy to read, and gave a good overview of the topic I requested. The references allowed me to check key facts and acquire additional information.

However, the citations were not always accurate. For example, in reference 11: Department of Homeland Security. (2024). Fiscal Year 2024 Budget Justification, here is the table that provides the protective function budget figures cited in the excerpt above. The “most recent fiscal year” budgetary figures cited for three Protective Operations categories ($907M, $82.5M, and $94.6M) match federal fiscal year 2023 in the table below extracted from the DHS document. However, federal fiscal year 2025 began in October 2024, and the most recent complete year is 2024, for which the figures do not match.  In addition, the report omitted the “Presidential Campaigns and National Security Special Events” category, even though its $73M budget is necessary to achieve the cited total 2023 budget of roughly $1.2B for Protective Operations.A document with text and numbers

AI-generated content may be incorrect.

Also, the section “Leveraging Technology for Enhanced Protection: The Role of Surveillance Drones” integrates information from several relevant references but fails to cite the 2023 “Privacy Impact Assessment for the United States Secret Service Unmanned Aircraft Systems Program (UAS)”, officially states how the Secret Service uses drones to aid its protective assets.  I found this reference via Google Search when researching this article.

In addition, the report also used or replicated an emotional or opinionated tone in some of its sentences, which could cast doubt on the objectivity of a human author. For example:

  • “The report meticulously details the security lapses…”
  • “…the Secret Service seeks individuals with strong ethics, morals, and virtues…”
  • “The Butler, Pennsylvania incident serves as a stark reminder…”

Language like this would be most appropriate for a report that advocates reforming the US Secret Service, but I did not ask for that. Of course, I could have asked Gemini to change the tone of the report, but what I got reflected the tone of the what it found on the web.

Also, Gemini behaves inconsistently. When I first generated the report, there were ample numbered citations, but the References section that identified the sources was missing. When I asked Gemini to add that section, it generated one that was missing some of the web references.

Furthermore, Gemini only finds references on the public web. As of this writing, it cannot accept uploaded documents.  However, as noted in this post from the AI Supremacy newsletter, Deep Research can be combined with Google’s Notebook LM to integrate generated reports with sources unavailable on the public web.

Conclusion

I found, during this and other similar experiments, Google Deep Research can assist enterprise and solutions architects in the discovery, integration, and analysis of public information. Therefore, enterprise and solutions architects will find it useful. However, per the footer displayed during every chat, “Gemini can make mistakes, so double-check it.”  

Authored by Iver Band, EA Principals Senior Instructor and ArchiMate Expert