Login
Register
Executives from Axiata Digital Labs, Deutsche Telekom, Telus and Vodafone see a role for the Open Digital Architecture (ODA) Canvas in scaling agentic AI.
Dawn Bushaus
DTW Ignite 2025: Telco execs highlight need for agentic AI standards
In a keynote panel discussion on Day 2 of DTW Ignite 2025, C-level executives from Axiata Digital Labs, Deutsche Telekom, Telus and Vodafone Group outlined a roadmap for generative and agentic AI in telecoms and called for enhancement of the ODA Canvas to scale AI.
The panel pointed to innovative ways they are using AI across their businesses, with a focus on improving sales and customer engagement, operational efficiency, and future innovation. But while many operators are implementing individual AI use cases, few if any are deploying AI at scale. Going forward, further standardization of the ODA Canvas to support scaling of GenAI and agentic AI use cases is required.
“How do you leverage all the APIs you have and actually construct really powerful end-to-end workflows?” said Telus’ Fahmy. “It’s one thing to have the APIs; it’s another thing say that this API is LLM ready.”
Telus is “heavily investing” in agent-to-agent protocol and model context protocol (MCP) servers to make internal assets consumable by large language models. MCP servers connect AI assistants to external tools, data and resources using the model context protocol – the idea is to let AI agents interact with networks more intelligently.
“I think this might be a challenge to the TM Forum: When we look at the ODA Canvas, what is the next evolution of it?” said Fahmy. “It’s not just ODA, but how is ODA LLM ready? Do we start packaging as MCP servers?” he added. “It will be interesting to look at the next evolution, because I think that’s where the real power comes – it comes out of the agentic AI.”
Axiata’s Rodrigo pointed to his company’s new Axonect Genix framework as a concrete approach to solving this challenge. The framework helps Axiata’s operating companies focus on two key priorities as the group’s IT strategy evolves: simplicity and velocity.
“We’ve been building an agentic framework on top of ODA to orchestrate what we already have,” Rodrigo explained. He illustrated the potential using an analogy of a fast-food restaurant, showing how AI can streamline service creation from intent to implementation.
In the images, an order-taker representing product management within the telco shouts orders at a fast-food cook (the CIO/IT team) who stands in the center of the room cooking on a grill (the ODA Canvas) with ingredients (thousands of APIs) laid out next to him.
“The cook makes this dish very quickly – you can see all his prepped ingredients on the side,” Rodrigo explains. And the interesting thing here is that it kind of reflects what we want to do: maximum speed and fuel simplicity.”
But there is a friction point between product management and IT, in that multiple subject matter experts today are involved in defining intent and implementing a solution. In the image, these are represented as zig-zagging lines. Axiata is introducing AI agents to simplify and speed the process.
“What we’ve been doing in the last 12 months is building this agentic framework on top of ODA to orchestrate what you already have, and more importantly to give this tool set to the product manager, who will now have a straight line to query with the agents what’s possible,” said Rodrigo, pointing to the image below to show the difference. “So, if you have an idea, you can actually find out what’s the shortest path to do it.”
Rodrigo adds: “It’s early days… We want to scale a bit during this year, and hopefully next year at this time we’ll have a lot more data points in terms of scale and usage and impact.”
Vodafone’s Stegmann agreed that more work to standardize the use of agents running on the ODA Canvas is needed and said Vodafone Greece is starting to work on solving these challenges as well. That operating company was the first to implement a commercial version of the ODA Canvas.
And at DTW this week, TM Forum members officially kicked off work on an AI-Native Blueprint, which will include best practices and standards for AI operations (AIOps), agentic AI, security and governance, and data architecture. The idea is for the blueprint to continuously evolve to enable AI-ready – and eventually AI-native – ODA.
But Stegmann concluded the panel with some words of caution: “The big challenge, I would say, for us at the end of the day is we’re running a business… We have to make sure that all the investments, all the effort that we put into our AI initiatives, really deliver value. And so, we must be careful that we don’t get all excited about the hype.”
Stegmann added: “We have to have serve our customers better. We have to open up new revenue streams. We have to really empower the industry to be more relevant in this AI future… That’s a question mark. What role are we going to continue to play here? Do we have a strong role to play? … These things cost money. It’s a big investment.”
