How AI-Assisted engineering improved ClimateView’s platform performance by 89%

Client_

ClimateView is a Swedish ClimateTech company that develops a SaaS platform designed to help cities, regions, ​and public authorities plan and execute their climate transition strategies.​

The platform enables CO₂e emissions analysis, scenario modeling, climate target tracking, and energy transition planning. It is used by 450+ municipalities and public-sector organizations worldwide to support data-driven decision-making and manage long-term emissions reduction programs.​

By combining climate data, action planning, and progress monitoring in a single environment, ClimateView helps public administrations turn climate ambitions into measurable initiatives and tangible outcomes.​

As part of the project delivered by Euvic, the team was responsible for the development and modernization ​
of the platform, focusing on improving data quality, enhancing application performance, modernizing the system architecture, and implementing an AI-assisted software development approach supported by Claude.​

Our role in the project_

The Euvic team was responsible for the development and modernization of key ClimateView platform components across both the backend and frontend layers. The scope of work included identifying and resolving performance bottlenecks, improving data quality and processing accuracy, redesigning the data import workflow, optimizing the application architecture, and increasing the solution’s testability and maintainability. ​

An important aspect of our contribution was the implementation of an AI-assisted software development approach, with Claude supporting the team throughout the engineering lifecycle. This included problem analysis, solution design, architectural decision-making, implementation, testing, and documentation of the development process. ​

Project goal_

The project focused on several strategic objectives related to data quality, platform performance, and the long-term scalability and maintainability ​of the solution. ​

Key priorities included: ​

• Data Reliability – Eliminating issues affecting the accuracy of emissions calculations. Improving the quality and consistency of data presented ​to users. ​
• Platform Performance – Reducing application response times. Enhancing ​the overall user experience. Mitigating challenges associated with processing large volumes of data. ​
• Modernization of Core Processes – Redesigning the data import workflow. Simplifying the solution architecture. Reducing technical debt that hindered further platform development. ​
• Quality and Change Safety – Increasing the solution’s testability. Reducing risks associated with deploying new features. Improving the predictability ​of large-scale architectural changes. ​
• Development Team Efficiency – Leveraging AI to support technical analysis and decision-making. Streamlining software design, implementation, ​and quality assurance processes. ​

Challenges_

As ClimateView continued to evolve, there was a growing need to enhance ​the reliability and scalability of the core mechanisms responsible for processing ​and analyzing climate data. For users relying on the platform to plan and manage energy transition initiatives, both the accuracy of results and the speed of access to information are critical. ​

​

Climate Data Reliability ​

One of the most significant challenges involved the calculation of emission factors ​for energy networks. The project involved a detailed review of the mechanisms responsible for emissions factor calculations, with a focus on improving consistency, reliability, and transparency across the data-processing workflow.​

Application Performance from the User Perspective ​

In selected areas of the platform, users experienced substantial delays when working with transformation-related components. Performance measurements revealed that the browser’s main thread was blocked for 12,407 ms, resulting in more than 12 seconds of waiting time before the application responded to user actions. ​

Complexity of Data Import Processes ​

As the platform matured, three independent data import mechanisms had been introduced. ​This approach complicated further development, increased the risk of inconsistencies, and made error handling and the implementation of new changes significantly more challenging. ​

Increasing Load on the Backend Layer ​

The analysis also identified issues related to inefficient processing of larger datasets and the occurrence of N+1 query patterns. In one process, generating an import summary alone required 17.14 seconds, with nearly 77% of the total execution time concentrated in just three operations. ​

Need to Reduce Technical Debt ​

Multiple parallel mechanisms performing similar tasks increased architectural complexity and hindered further platform development. The challenge was therefore not only to address current issues but also to establish a solid foundation that would enable the platform to evolve safely and efficiently ​in future development phases. ​

Workflow_

ClimateView’s development followed an iterative approach based on data analysis, hypothesis validation, and continuous measurement of implemented improvements. In most cases, the goal was not simply to eliminate visible symptoms, but to identify the root cause of each challenge and select a solution that would support the platform’s long-term evolution. Data-driven decision-making, performance metrics, and the use of Claude as an engineering support tool played a key role throughout this process. ​

Step 1. Problem Analysis and Root Cause Identification ​

Each initiative began with a thorough understanding of the issue and the identification of its potential causes. The team analyzed application behavior, ​solution architecture, and the data processed by the platform. At this stage, Claude supported repository reviews, dependency analysis between components, and the identification of potential sources of issues. ​

Step 2. Hypothesis Validation ​

The next step involved validating assumptions against the actual behavior ​of the system. Hypotheses were tested using performance measurements, operational data, and observations gathered during application usage. ​For more complex challenges, unsuccessful investigation paths were also documented, ensuring full analytical context and preventing the team from revisiting the same dead ends. ​

Step 3. Solution Design ​

Once the root cause had been identified, the team evaluated multiple solution approaches. The assessment considered the impact of proposed changes on system architecture, code maintainability, performance, and future product development. Claude supported the comparison of available options and helped identify potential technical trade-offs before implementation work began. ​

Step 4. Implementation and Testing ​

Development work started only after the analytical phase had been completed. Activities included backend and frontend implementation, test development, and the gradual validation of new functionality. Claude was used to support tasks such as unit test creation, edge-case identification, and refactoring efforts. ​

Step 5. Results Measurement ​

Every optimization was evaluated using clearly defined technical metrics. Measurements were performed both before and after implementation, making it possible to accurately assess the impact of changes on the platform’s performance, stability, and overall quality.​

Strategy adopted_

The foundation of the adopted strategy was the use of Claude not as a code-generation tool, but as a support system for engineering processes ​and technical decision-making. The team treated AI as a collaborative partner that helped analyze complex system dependencies, evaluate alternative approaches, and maintain the context of ongoing investigations. In practice, Claude supported several key areas of work. ​

Root Cause Analysis ​

The most complex challenges required multi-stage investigations involving the validation of successive hypotheses. Claude supported the process of identifying the true root causes of issues by helping organize findings, document the course of investigations, and analyze dependencies between system components. This approach made it possible to preserve the full analytical context, even during investigations that spanned several days. ​

Evaluation of Solution Alternatives ​

Before implementation began, the team compared alternative approaches based on their impact on architecture, system maintainability, cost of change, and future product evolution. Claude supported the evaluation of available options and highlighted potential trade-offs that needed to be considered during the decision-making process. ​

Decomposition of Large Initiatives ​

For complex modernization efforts, the strategy focused on breaking ​large initiatives into smaller, independent phases that could be implemented incrementally. This approach reduced the risks associated with large-scale refactoring efforts while enabling improvements to be delivered more quickly, without waiting for the completion of an entire transformation program. ​

Supporting Quality and the Development Process ​

Claude was also used during architecture analysis, documentation creation, edge-case identification, code reviews, and unit test preparation. As a result, AI supported not only the implementation phase but the entire software development and platform maintenance lifecycle. ​

Delivered solution_

As part of the project, a comprehensive modernization of key areas of the ClimateView platform was carried out. The changes addressed data quality, system performance, ​and the long-term evolution of the platform’s architecture. The work focused on four main areas. ​

​

More Reliable Climate Data ​

We modernized the mechanisms responsible for emissions calculations, strengthening the consistency and reliability of data processing. This provides stakeholders with a stronger foundation for planning, monitoring, and evaluating climate transition initiatives.​

Modernized Data Import Process ​

We redesigned the data import workflow to make it more robust, maintainable, and easier to extend. The new approach provides greater operational predictability, improved visibility into data quality issues, and more efficient handling of large datasets. ​

Faster Data Processing ​

A series of backend optimizations was implemented to accelerate data processing ​and reduce the execution time of the most resource-intensive operations. ​These improvements increased the platform’s ability to efficiently manage larger ​data volumes, a critical capability for cities and local governments performing complex climate and energy-related analyses. ​

More Responsive User Experience ​

We optimized the frontend application by removing bottlenecks that were causing delays during user interactions. As a result, the platform delivers a smoother experience, enabling faster and more comfortable access to critical information. ​

​

The ClimateView project demonstrates how artificial intelligence can effectively support the resolution of complex engineering challenges far beyond code generation. Throughout the project, Claude was used as a tool to augment the work of experts in system architecture analysis, code dependency identification, decomposition of complex initiatives, as well as test and documentation preparation. ​

As a result, the team was able to evaluate potential scenarios more efficiently, validate hypotheses faster, and focus on making decisions with the greatest business impact. AI did not replace the knowledge and expertise of engineers; rather, it served as an additional layer of support that helped improve delivery efficiency and the quality of the solutions provided. ​

Benefits_

Higher Data Quality and Reliability ​

The project enhanced data processing mechanisms and strengthened the consistency and reliability of calculations used by the platform. This provides a robust foundation for emissions-related analysis, reporting, and business decision-making.​

Significant Performance Improvements ​

Optimization of the data import process reduced main-thread blocking time by approximately 89%. The result is a more responsive platform, ​capable of efficiently handling growing volumes of operations and data. ​

Architecture Ready for Future Growth ​

The redesign of the data import process and the consolidation of data processing mechanisms simplified the overall solution architecture. ​This made future feature development easier, faster, and more predictable. ​

Controlled Change Delivery ​

The modernization effort was divided into 10 independent phases, enabling incremental delivery of improvements while maintaining uninterrupted platform operations. This approach allowed the team to continuously validate outcomes and reduce the risks associated with large-scale system modifications. ​

Greater Stability and Solution Quality ​

Separating business logic from infrastructure components improved the ability to test critical processes and streamlined application maintenance. ​As a result, the organization gained a more stable solution that is easier to maintain and evolve. ​

Faster and More Responsive Application ​

Optimizations to the data import process resulted in quicker interface response times and a smoother user experience, particularly when working ​
with larger datasets. ​

More Predictable Processes ​

Improvements to data processing mechanisms made imports more structured and predictable, helping users perform their daily tasks more efficiently. ​

Improved User Experience ​

Reducing application workload minimized the risk of slowdowns during critical operations, increasing overall usability and user comfort. ​

Technologies_

• AI: Claude ​
• Backend: Glang (Go), Fiber v2, GORM, sqlc, gRPC ​
• Frontend: Angular, AG Grid, PrimeNG, Angular CDK, Tolgee ​
• Data & integrations: Excelize, tus, Swagger ​
• Performance/Observability: OpenTelemetry, Chrome Performance ​
• Delivery/Project Management: CI/CD, Linear ​

Key numbers_

89%​ – Reduction in application main-thread blocking time​

12 407 ms → 1 384-1 410 ms ​– Reduction in user wait time for an application response, from over 12 seconds to approximately 1.4 seconds 

10 implementation stages​ – Used to divide the data import redesign into independent initiatives that could be deployed safely​​

34 code locations analyzed​ – As part of an audit of potential application performance bottlenecks​

Summary_

The project involved redesigning the mechanisms responsible for emissions calculations, modernizing the data import process, optimizing the processing of large datasets, and reducing technical debt that was hindering further system enhancements. A particularly important aspect of the initiative was the use of Claude as a support tool within the engineering process, primarily for analyzing complex challenges, evaluating architectural alternatives, and planning modernization efforts. These improvements resulted in a more stable platform, faster application performance, and a more predictable environment for the product’s ongoing development and future growth. ​