Success Story: Seguas

Zaragoza, Reus, Bilbao

902 02 62 13

Intelligent Maintenance Assistant for SEGUAS on AWS

SEGUASA company specializing in compressed air, industrial refrigeration, and comprehensive maintenance needed to transform how its technicians accessed technical information. With over 5.000 industrial assets under management, documentation was scattered across manuals, work orders, maintenance records, and various internal systems.

AI Assistant

What challenge did the company face?

SEGUAS needed to reduce the time its technicians spent searching for technical information and improve the efficiency of its maintenance operations.

AI Assistant for industrial maintenance

The SEGUAS technical team faced several operational limitations:

Scattered and unstructured information
Technical manuals, maintenance protocols, work orders, and historical data were distributed across multiple sources such as SharePoint, Odoo, and internal repositories.

Difficulty finding quick answers
The technicians spent too much time searching for documentation before they could diagnose or resolve an issue.

Dependence on expert knowledge
Problem-solving depended largely on the individual experience of each technician.

Large volume of documents
The system had to manage more than 10.000 technical documents and more than 100.000 pages, making manual processing unfeasible.

Need to organize the data
Before applying AI, it was necessary to improve the structure, quality, and metadata of the available information.

Solution implemented

Apser designed and implemented on AWS a intelligent maintenance assistant, capable of allowing technicians to query information in natural language and obtain contextualized answers about equipment, work orders and technical documentation.

Main components of the solution

Intelligent conversational assistant
Technicians can consult the system in natural language to obtain technical information, indicative diagnoses, and recommendations based on real documentation.

RAG architecture on AWS
The solution uses Amazon Bedrock and a Retrieval-Augmented Generation approach to generate answers based on reliable documentation stored in Amazon S3.

Integration with corporate data sources
The wizard connects information from SharePoint, Odoo, and internal repositories, unifying access to previously scattered data.

Agentic model with backend tools
The assistant dynamically selects which action to perform based on the user's query: retrieve work orders, identify associated machines, consult manuals, or review maintenance history.

Operational data query
Amazon Athena allows you to query operational data related to work orders and equipment, while Amazon DynamoDB stores structured solution information.

Scalable processing on AWS
AWS Lambda and containerized backend services allow for scalable processing of events, queries, and interaction flows.

Results

The solution enabled SEGUAS to transform access to technical knowledge and improve the efficiency of its maintenance operations.

Faster access to information
Technicians can locate relevant documentation, history, and data from a single conversational interface.

Improvement of technical diagnostics
The assistant provides contextualized answers and guides the technician during incident analysis.

Less dependence on individual knowledge
Technical knowledge becomes more accessible to the entire organization, reducing dependence on specific experts.

single source of knowledge
The solution centralizes information from multiple systems and facilitates more consistent access to data.

Better experience for technicians
Operators can consult necessary documentation even from the client's premises, with an experience similar to what they would have from SEGUAS' offices.

Lessons learned

Data quality is key

The accuracy of the assistant depends directly on the structure, quality, and consistency of the documentation.

Metadata makes the difference
Correctly linking manuals, machines, work orders, and operational data is essential to obtaining useful answers.

The improvement is iterative
Adjusting prompts, recovery strategies, and agent logic was necessary to improve accuracy and reduce incomplete responses.

Agentic AI improves usability
Moving from a static search to an assistant capable of guiding workflows improves usability in real-world maintenance scenarios.

Digital transformation starts with people.
The success of the project required training, a change in habits, and the adoption of a data-driven approach.

Why Apser

apser is the AWS Advanced Consulting Partner specialized in private companies, public sector, ISVs and non-profit organizations.

The apser team provides a customized, agile and understandable approach that contributes to the widespread adoption of the cloud paradigm.

In 2018 we began our experience with AWS and, since then, we have not stopped incorporating knowledge and experience, a fact that drives us to improve the efficiency of our clients' platforms.

Our team operates through two business units: Cloud Solutions (as a managed services provider) and Cloud Professional Services (as consultants).