AI Framework speeds productivity

It all started with a provocation: “how can we apply AI to software development?”. From this, the CWI team, which acts as a service provider at BV bank, took on the mission of creating a framework that guides how and in which scenarios to use an AI tool to increase productivity in software development.

First Steps

Initial studies were carried out based on the available possibilities and a framework was created focusing on the use of this tool. However, in early 2024, a new chat feature with a more advanced model was implemented.

Growth Space

Quickly, the set of practices and tools was adapted to make the most of the new developments. With this evolution, the documentation was revised and transformed into a page on the CWI GitLab wiki, containing various materials that increased productivity in the work routine.

In addition to guiding how and in which scenarios to use the AI tool, the framework reinforces security issues, explains limitations, exemplifies which prompts can be used, provides guidelines for generating unit tests, API contracts, creating DDL and queries, and also acceptance and exception scenarios for the quality team.

Cutout of a man's hands working on a notebook, with AI symbols added on top.

Real Results

To measure the practical gain from the framework, a study of metrics was carried out on some tasks, including unit tests.
The strategy created uses labels in Jira: whenever a developer creates a unit test, he adds a label to the story under development. In the Java context, this guarantees the quality of delivery, allowing the identification of rule changes and necessary adjustments in tests. The developer enters code with a prompt and the AI tool returns an applicable unit test example, speeding delivery.
Currently, the focus is on expanding the framework to increase profits, working on materials for proof-of-concept scenarios, code conversion and explanation, query optimization, Regex generation, support in analyzing problems in production, unit testing for Angular Front-End and automated testing.

A tablet displays an illustrative graph of revenue over a period of time.

Continuos Evolution

Artificial Intelligence, although it is the main topic addressed within CWI’s AI area, is being integrated into other quadrants as the range of opportunities expands.

This work reflects CWI’s culture of continuous learning and a lot of knowledge exchange. Each presentation about the framework brings new insights that are incorporated into the project, creating a cycle of learning and improvement. The integration and involvement of people are essential to awaken new ideas and motivations.

Two men talk in an environment with white walls, with a window in the background.

Circular Knowledge

One of the next steps involves disseminating the knowledge acquired, through workshops and immersions. The objective is to share this knowledge in a compiled form.

We are proud to have been pioneers in approaching this framework, and we believe that the work reflects the way CWI operates: solving complex problems in a simplified way, bringing real benefits to the partner.