Dynata AI scale throughout the development cycle with AWS AI-DLC and Kiro IDE

Dynata, a global data provider for market research and advertising, started applying the AI-DLC (AI-Driven Development Life Cycle) approach with AWS's Kiro IDE to scale the use of AI at all stages of the software development cycle. According to the case published in the AWS architecture blog, the results in code generation were positive from the outset, with productivity gain and quality maintenance. The company also connected assistants to its own internal knowledge ecosystem.

What is AWS AI-DLC and what changes in development?

AI-DLC is the AWS methodology that reorganizes the software life cycle — requirements, design, implementation, testing and operation — around AI agents, with the developer acting less as a typist and more as a reviewer and orchestrator. In practice, this displaces the bottleneck: the work ceases to be writing line by line and becomes to define context, validate decisions and check what the agent produced. It is the same logic that AWS has been promoting in regulated sectors, as in the AI-oriented life cycle for financial applications.

Why did Dynata choose the AWS IDE Kiro?

Kiro IDE is AWS's agentic development environment that works from specifications: before generating code, the agent produces requirements documents, technical design and a list of tasks that work as a contract for implementation. This "spec before code" format is what gives predictability to the process and facilitates human review — a central point for a data company that needs to maintain traceability and compliance. The detail that the case highlights is the maturation time: the positive results appeared from the beginning, without a long adjustment phase of the tool.

How does the AI-DLC connect to the company's knowledge?

The second movement described in the case is the most relevant for those who have tested code assistants. Dynata has linked the AI to its internal knowledge ecosystem, so that the agent stops responding only with generic standards and starts to consider architecture, conventions and the organization's own history. It's the difference between an assistant who writes plausible code and one who writes code that fits the system that already exists. Self-complete tools rarely solve this second problem, because they do not see the context of the company.

What Dynata's case with AI-DLC doesn't say

The material does not disclose percentages of productivity, volume of code generated, size of the team involved, deployment time or licensing costs. There is also no comparison with approaches that Dynata has tested before. Without these numbers, it is possible to state that the initiative worked in the company, but not estimate the return in another organization.

What does AI-DLC and Kiro IDE change to devs and companies in Brazil?

Those who develop in Brazil already have access to agentic environments such as Kiro and methodologies derived from AI-DLC, so the entry barrier is no longer the tool. The bottleneck is organizational: a tidy knowledge base, governance about what the agent can change, code review and standardization between teams. It is also what shows surveys on AI returns in companies, such as IDC study on real ROI of AI: the gain tends to appear where there is process, not just license. Brazilian companies that treat code AI as an isolated shortcut tend to reap less than those that integrate it in the entire cycle.

Frequently Asked Questions about AWS AI-DLC and Kiro IDE

What is AI-DLC?

It is the acronym for AWS' AI-Driven Development Life Cycle, which reorganizes the stages of software development — requirements, design, code, testing and operation — around people-supervised AI agents.

What is Kiro IDE?

It is the agentic development environment of AWS, which generates requirements and design specifications before writing code, using these documents as a reference for implementation.

Does Dynata's case prove productivity gain with AI?

The case reports productivity gains and quality maintenance from the beginning, but does not publish closed percentages or metrics, which prevents generalizing the result for other companies.