top of page
TIP_horizontal_rgb_white.png

From rApp Portability to Agentic RAN: Building Open Network Automation

thumbnail showing Signal Boost participants.

As networks become more software-driven, the telecom industry is entering a new phase of automation. For Open RAN and brownfield networks alike, the question is no longer whether AI and machine learning will play a role in network operations. The question is how quickly the industry can make intelligent automation practical, portable, validated, and ready for real-world deployment.

In this episode of The Signal Boost, TIP’s Chief Commercial Officer Vishal Mathur is joined by Abdel Bagegni, Head of OpenRAN Technology at TIP, and Michael Behan, Solutions Architect at Zinkworks, for a practical conversation on rApps, SMO/RIC readiness, and the future of AI-driven RAN automation.

The discussion comes at an important moment for the industry. Operators are looking for ways to improve network performance, reduce operational complexity, lower energy consumption, and prepare for more autonomous network operations. rApps, non-real-time RICs, and SMO frameworks are becoming a key part of that evolution, but only if they can move beyond isolated proofs of concept and into scalable, interoperable deployment.

rApp portability as a real industry challenge

Within TIP’s OpenRAN Project Group, the RAN Intelligence, Automation and Orchestration subgroup is focused on translating operator requirements into practical use cases, technical specifications, validation frameworks, and test plans. Abdel explains that the group’s work is centered on AI/ML-driven automation, with rApp portability as a core objective.

That matters because operators do not want intelligent network applications that only work in one environment, on one vendor’s stack, or under one narrow set of conditions. The original promise of Open RAN was greater choice, flexibility, and interoperability. If the next generation of rApps and automation tools creates another layer of vendor lock-in, the industry risks moving the problem up the stack instead of solving it.

TIP’s work is designed to help avoid that outcome by creating vendor-agnostic requirements and test frameworks that can support broader adoption. Since FYUZ in Dublin, TIP members and operators have been working through an SMO/RIC assessment and development framework, including commercially relevant use cases such as geolocation, anomaly detection, root cause analysis, and traffic steering.

The goal is not just to define what rApps could do. It is to help the ecosystem understand how they can be tested, validated, onboarded, and deployed in a way that gives operators confidence.

Moving beyond the prototype

For Zinkworks, the challenge is not simply building rApps. As Michael explains in the episode, building an rApp is only one part of the problem. The larger challenge is everything that happens after the prototype: training it on real network patterns, validating behavior before it reaches production, deploying it across different SMO environments, monitoring performance, governing updates, and rolling it back if something unexpected happens.

Zinkworks has built its approach around five pillars: build, train, validate, deploy, and govern.

That full lifecycle view is critical because intelligent RAN applications will increasingly influence live network behavior. An energy-saving rApp, for example, cannot simply shut down capacity during low-traffic periods without understanding handover behavior, demand patterns, customer experience, and service commitments. A root cause analysis or anomaly detection rApp needs the right data, the right validation, and the right operational guardrails before it can be trusted in a live environment.

This is where the industry has to move from “can we build it?” to “can we safely and repeatably deploy it?”

Vodafone and Zinkworks: a practical proof point

That shift is already taking shape through Zinkworks’ partnership with Vodafone on Rapid RIC, a Generative AI-driven platform designed to help Vodafone engineers develop, deploy, validate, and monitor rApps across multiple markets.

Vodafone has described Rapid RIC as a central platform that combines secure data analytics, an intuitive visual interface, and code-generating AI to help engineers build network software applications more quickly. The platform is expected to be fully operational by early 2026 and is intended to support the launch of customer-focused RAN applications within weeks rather than months.

The use cases are directly tied to operational value: improving network quality, automatically detecting and resolving faults, reducing energy use, saving costs, and improving signal quality in both high-demand and rural areas. Vodafone expects the platform to reduce RAN application development and launch times by 60 to 70 percent, with new features deployed in 3 to 4 weeks instead of months.

For the broader ecosystem, this is a useful example because it shows the direction of travel. The opportunity is not just faster app development. It is a more repeatable model for building, validating, and operating intelligent network applications at scale.

Why real network data matters

One of the harder questions in AI-driven network automation is how to train applications on the right data. Synthetic data has an important role to play, particularly for testing and simulation, but models intended for live network environments need to reflect real network behavior.

That creates a practical challenge. Operator network data is highly sensitive, and it is not easy to share across vendors, developers, or test environments. Abdel points to active work inside TIP to create templates, clusters, and frameworks that could help make domain-specific network data usable in a trusted way.

This connects directly to TIP’s broader work around Data & AI Foundations, where the community is exploring how to abstract the value of operator network data and expose it through network language models, federated learning techniques, and secure sandbox environments.

For rApps, this matters because the value is only as strong as the confidence operators have in the application. Better data, stronger validation, and clearer governance are what make intelligent automation usable in real networks.

Energy saving as a first step on the autonomy ladder

Energy saving is one of the most immediate and practical use cases discussed in the episode. The RAN accounts for a significant share of mobile network energy consumption, and operators have long explored ways to switch off capacity layers during low-traffic periods.

What is changing now is the intelligence required to do it well.

AI/ML-driven rApps can help predict traffic demand, manage handover implications, and balance energy savings against quality of service. But this is exactly the kind of use case that needs rigorous lifecycle management. If an energy-saving application causes dropped calls or degraded service when traffic patterns shift, the operational risk quickly outweighs the energy benefit.

That is why validation, monitoring, and rollback mechanisms are not optional. They are what turn automation into something operators can trust.

From Open RAN automation to agentic intelligence

The episode also looks ahead to the role of agentic AI in network operations. Michael describes an emerging layer of intelligence that could reason across rApps, compose capabilities, adjust parameters, and make trade-offs based on operator intent.

In that model, the SMO is not replaced. It is elevated.

The agentic layer would sit above or across the traditional automation environment, helping coordinate decisions across applications and use cases. But this future depends on the foundations being built now. An agentic system that can autonomously create or deploy rApps still needs strong governance, validation, SMO-agnostic design, and clear operational guardrails underneath it.

In other words, the industry cannot skip the hard work of lifecycle management. The path to agentic networks runs through practical, validated, interoperable automation.

What comes next

As the industry looks toward 6G, AI-native networks, and more autonomous operations, the work happening around rApp portability and SMO/RIC frameworks is becoming increasingly important. Operators need a way to capture the benefits of AI-driven automation without sacrificing openness, interoperability, or control.

That is where TIP’s community model plays an important role. By bringing operators, vendors, and technology partners together around shared requirements, test plans, and deployment-focused use cases, the ecosystem can move faster while reducing fragmentation.

Zinkworks is one of the companies helping push this work forward, both through its participation in the Open RAN ecosystem and through practical deployments like Rapid RIC with Vodafone. Listen: https://open.spotify.com/episode/3Py51AM5KnXOjBxmoHyRGW?si=SX9DFWXQSsqn-2cSLyuj_Q


Continue the conversation at FYUZ

This conversation will continue at FYUZ in Seattle, where TIP will bring together operators, vendors, system integrators, policymakers, and technology leaders to focus on the real-world evolution of telecom infrastructure.

As a FYUZ sponsor, Zinkworks will be part of the broader discussion on Open RAN, AI-driven automation, rApp portability, and the move toward more intelligent, software-defined networks.

Join us at FYUZ to connect with the companies and communities shaping what comes next for open and intelligent network infrastructure.


bottom of page