Revolutionizing a 100-Year-Old Industry with AI-First Strategy

Secil is a business group founded in Portugal whose activity is based on the production and sale of cement, concrete, aggregates, mortars and hydraulic lime. It also integrates companies that operate in complementary areas in the circular economy, in the use of waste as a source of energy.

The company embraced an AI-first strategy as the cornerstone of its digital transformation. AI now plays a pivotal role across nearly every function—from architectural design to logistics and supply chain management. While MuleSoft has long served as the primary integration platform, the increasing demands of a growing project backlog and a lean development team highlighted the need to go beyond traditional development methods to maintain a competitive edge.

“No one wants to do code reviews or MUnits. CurieTech AI removed that emotional stress from the team.”
_{Guilherme Pinho, Solution Architect and Lead Secil}

Challenges

Secil’s integration teams support seven business units, each generating a steady flow of discovery-based initiatives. Most existing processes were still paper-based, and the increasing demand exceeded the current technical capacity. Key challenges included:

• Small team, big backlog
  With only four developers, Secil’s integration team was overwhelmed by requests. Critical tasks like code reviews and test case creation consumed a disproportionate amount of their time.
• Generic LLMs missed the mark
  Initial tests with mainstream LLMs didn't produce value. These tools weren’t trained on MuleSoft’s specific syntax and logic, which made them unreliable for specialized tasks.
• Manual code review bottlenecks
  Code reviews accounted for roughly 30% of the team’s cycle time—mostly due to delays waiting for peer availability.
• Time consuming testing process
  MUnit creation for complex process APIs consumed significant time, especially considering the team’s internal requirement of maintaining at least 80% test coverage.

The team sought to streamline development and enable faster delivery with modern AI tools.

Why CurieTech AI?

Secil found CurieTech AI through a Google search and immediately saw the difference. Unlike generic models, Curie’s domain-specific AI agents are trained for MuleSoft—from DataWeave to Code Review to Munit generators. CurieTech AI stood out for several reasons:

• Code Review agent that acts as a senior developer
  CurieTech AI agents replaced manual reviews, reducing wait times to almost zero while maintaining high quality.
• 50% combined time savings with testing and code reviews
  CurieTech’s MUnit and Code Review agents cut code validation time by over half, helping the team meet delivery goals and take on additional projects without extra hires.
• Zero developer burnout on mundane tasks
  Mundane tasks like test writing and code review, often seen as tedious and demotivating, were automated—allowing developers to focus on high value tasks.
• Frees developers to focus on what inspires them
  CurieTech AI agents handle tasks like code reviews and testing, acting as senior developer support. This allows teams to focus on solution design, architecture, and strategy—the work they’re most passionate about.

“Having Curie was like adding another senior developer to our team of four.”
_{Guilherme Pinho, Solution Architect and Lead Secil}

Benefits and Outcomes

Since adopting CurieTech AI, the team at Secil has experienced significant efficiency gains:

- 30%+ reduction in total development cycle time
- Code review wait time dropped to nearly zero
- 50% time savings in MUnit generation and validation
- Increased team satisfaction and focus on higher-value design work
- Ongoing support and engaging collaboration with CurieTech AI team

“The worst thing we could do is not try AI. If we wait, we risk becoming obsolete.”
_{Guilherme Pinho, Solution Architect and Lead Secil}

Next Steps

Secil plans to expand its use of CurieTech AI across more workflows and integrate it more deeply into their daily development cycle.

“For MuleSoft users, generic AI tools won’t cut it. But with domain-specific agents, the ROI is clear. Just start.””
_{Guilherme Pinho, Solution Architect and Lead Secil}