logo_header
  • Topics
  • Research & Analysis
  • Features & Opinion
  • Webinars & Podcasts
  • Videos
  • Event videos

Axiata combines GenAI with the ODA Canvas to accelerate development

Axiata Group’s ambition to become a techco has led it from an API-first approach to the adoption of ODA and the ODA Canvas. Now it is going a step further and combining GenAI with its implementation of TM Forum’s ODA Canvas.

Joanne TaaffeJoanne Taaffe, TM Forum
06 Nov 2024
Axiata combines GenAI with the ODA Canvas to accelerate development

Axiata combines GenAI with the ODA Canvas to accelerate development

The Axiata Group foresees making considerable savings on solution development time by combining GenAI with its implementation of TM Forum’s Open Digital Architecture (ODA) Canvas, according to Namal Jayathilake, Chief Technology Officer, Axiata Digital Labs, speaking at TM Forum’s Innovate Asia event.

The Axiata Group has been using TM Forum’s Open APIs to expose its IT core capabilities for use internally, as well as externally by partners and enterprises since 2018. It has also built on its ‘API first’ approach by adopting TM Forum’s Open Digital Architecture (ODA) and the ODA Canvas to ”develop a platform … to get the best of ODA implemented in a real-life scenario platform as a middleware,” explained Jayathilake.

This has allowed Axiata to give operators within the Group a “common fabric layer which seamlessly implements standardization, observability, security, automation and many other functionalities …. [demanded by] modern day micro service or solution creation.”

At the same time Axiata has been careful to establish independent run times for each of the core capabilities so that it can ensure common governance while allowing each operating company to independently create and use components.

“While keeping the component functionality intact within the micro service …you create on top of the Canvas. And this is naturally API driven with common operations, module architect and component reusability and so on,” explained Jayathilake.

In addition the company has created templates of 50 TM Forum APIs, ODA components and business cases for its implementation of the ODA Canvas. Pre-built components means the company “can gain speed, and promote the reusability … [and] standardization,” said Jayathilake. “There's no limitation, and now we are basically replicating the same success story to CSPs” within the Group.

Linking GenAI to ODA Canvas

Now Axiata is experimenting with harnessing Gen AI to further accelerate development and implementation.

The Group has developed a GenAI based module made up of two components; the design studio for solution engineers and a studio for DevOps engineers, running on top of the Canvas implementation.

This allows Axiata to “upload a BRD [business requirements document], generate the micro service structure of the solution… the code … [and] the unit test cases. We can execute the unit test cases and we can deploy [it] and we can do bit of operational management as well,” according to Jayathilake.

Initial implementations and trials show that Axiata “can save around 30% [of time] at the business architecture and business requirement creation level” and 20% of the time needed for design, through the generation of the micro service structures and solution architectures, said Jayathilake.

“You can drag and drop things, regenerate certain designs, generate API signatures and so on."

The Group is also reducing the time spent during the continuous testing and continuous deployment phase by automating case writing and test case execution.

“You can generate test cases aligned to BRDs and then execute some of these things at least at the API level or the user level and you save a lot of time. And when you road test with modules created by Gen AI at the API level, again, you save time."

The company is also looking to extend the combined use to Gen AI and the ODA Canvas to business and project management tools such as Jira.

“For example, you can record your team's code, convert it to a text file, upload the BRD, generate the test cases, test, deploy,” said Jayathilake. “You just need the initial discussion to understand the requirement in detail and then deploy. That's … the value we are heading towards, and it looks very promising.”