Bromma: Leading service operations with AI-powered search and automation

Bromma is a global leader in manufacturing crane spreaders for harbors and port terminals. Their success isn’t just about building high-quality machinery – it’s about ensuring seamless support when service issues arise. But outdated systems were making that increasingly difficult.

Their internal system for logging and diagnosing quality issues, dating back to the early 2000s, was both complex and inefficient. Information was fragmented, with critical service details and history located in unstructured data fields in the legacy database. The system’s limited search capabilities meant that engineers had to manually search through past reports, relying on their experience to find relevant solutions.

To tackle these inefficiencies, Bromma partnered with Futurice and Futurice Family data company Recordly to implement a retrieval-augmented generation (RAG) agent – a game-changing generative AI solution designed to streamline service operations.

To ensure seamless interaction, the solution includes a user-friendly interface, allowing service engineers to engage directly with the AI agent through natural conversation. This intuitive design ensures that engineers can quickly access the insights they need, improving efficiency and decision-making.

By leveraging natural language processing and intelligent search, the agent processes Bromma’s quality service reports (QSRs) and extracts relevant data points, answering key service questions instantly, such as “What was the issue reported in this QSR? Was the issue resolved? If yes, what was the solution?”

The data pipelines extract and process the data into a format the agent can read and ensure that engineers can now use simple, natural-language queries to retrieve the most relevant cases. Instead of manually searching through legacy reports, now they can simply ask “Have we encountered issue X before?” or “How do I resolve issue X?”

The agent doesn’t just retrieve data - but it also generates step-by-step diagnostic instructions and on-site testing procedures, helping engineers resolve issues faster and more effectively.

Built on Google Cloud Platform, the solution harnesses modern technologies and language models to power its operations. It integrates Google BigQuery, Google Dataform, and Google Cloud Run to efficiently process, store, and serve data in the correct format for the RAG agent.

The solution leverages Google Gemini as the primary LLM for the user interface, while multiple models operate in the background to retrieve documents and rank search results based on relevancy.

Users are guided through the functionalities of the RAG agent with built-in prompting instructions within the application. Additionally, the agent provides curated suggestions when user inputs are too vague. The user can also be prompted to filter the search conditions of the search.

The AI-powered agent was recently rolled out successfully, and while the long-term ROI is still being measured, early feedback from key users – including IT stakeholders and service engineers – has been overwhelmingly positive.

By eliminating tedious manual searches and empowering engineers with fast, AI-driven, insights, Bromma has transformed its service operations - saving time, reducing downtime, and ensuring a more seamless customer service experience.

About Bromma

For over 50 years, Bromma has delivered crane spreaders to more than 500 terminals in over 90 countries on six continents and is the global market leader and most experienced manufacturer in its business. In fact, in any port you visit, you’re likely to see a Bromma spreader in action. The company is a part of Kalmar.