Scrum Case
How the FBI Developed in 1 year a project that was 10 years late with 90% cost savings?
6 de fevereiro de 2020
By Luciano Osorio – Scrum Master at Zappts
Free Translation and Adaptation of the original text Software in 30 Days, written by Ken Schwaber and Jeff Sutherland (creators of Scrum).
In 2003, the FBI decided to digitize the case files of their investigations. This would allow investigators to quickly compare cases and discover connections between them.
The project for automating the comparison processes was called Sentinel.
Sentinel Before Scrum
In March 2006, the FBI began developing Sentinel. The projection was a user base of more than 30,000 end users, including agents, analysts, and FBI administrative staff. Original estimates were US$451 million for the development and implementation of Sentinel by December 2009.
According to the FBI’s original plan, Sentinel was to be developed in four phases. The FBI hired security firm Lockheed Martin for the project. This company proposed using the traditional software development methodology called Waterfall.
In August 2010, the FBI spent US$405 million of the US$451 million Sentinel budget. However, the FBI delivered functionalities for only two of the four phases of the project.
Even though these deliveries improved the FBI’s case management system, they did not generate as much value as predicted. Due to cost overruns and a tight schedule, the FBI canceled Sentinel in July 2010, leaving it incomplete.
Scrum and the New Project Approach
The FBI decided to resume the project with a new methodology. The intention was to see if there would be improvement in results. This new methodology is called Scrum.
According to the Standish Group CHAOS Report, only 37% of software projects are successful. Only 14% of traditional (Waterfall) projects succeeded, compared to 42% of agile projects (Scrum).
Beyond the Standish Group’s traditional definitions of success, these projects also allowed greater responsiveness to changes in customer needs, enabled better risk mitigation, and finally provided higher quality software.
In 2009, the FBI recruited a new Chief Information Officer (CIO) and Chief Technology Officer (CTO) with experience managing organizations that built software using Scrum.
In 2010, the CTO decided to change the Sentinel approach. This would simplify decision-making processes and allow the FBI to deliver Sentinel within budget.
The FBI told the Inspector General of the Department of Justice that it believed it could complete Sentinel with the remaining budget and within 12 months.
An audit had already concluded that the FBI would need an additional US$35 million and six more years if it had continued with the traditional methodology.
The FBI reduced the Sentinel team from 400 to 45 people, 15 of whom were programmers. The CTO led the project alone, with the goal of delivering new Sentinel functionalities every 30 days. Each new functionality needed to meet all final requirements.
Every three months, the FBI would implement the features built in the three preceding iterations in a field pilot.
Sentinel One Year Later
In November 2011, after one year of restarting the project with Scrum, all phases of Sentinel were completed.
The software was implemented in a pilot group of FBI offices. The remaining offices had implementation by June 2012. The FBI completed Sentinel for US$30 million in 12 months, a savings of more than 90%.
After the FBI changed its approach, adopting Scrum as a Digital Transformation tool, they worked with the same intensity as before but were rewarded with much superior results.
If an organization like the FBI can do it, why can’t yours?
Share this article
Related articles
16 set 2026
The End of Passive SaaS: Why You’ll Pay for Outcomes, Not Seats
The traditional software pricing model based on per-user licenses (seat-based SaaS) faces an inevitable decline in 2026.
09 set 2026
The Timid Autonomy Dilemma: Why Keeping AI in a Suggestion-Only Role Is Killing Your Margins
This article analyzes the financial impact of this "timid autonomy" and advocates for an urgent shift to the "Human-on-the-loop" (HOTL) model.
02 set 2026
The "SaaSocalypse" is actually an architecture and identity crisis.
This article reverse-engineers a real-world success story (anonymized) from the financial sector, dissecting the layers of...