Qiang Li

🧨 OpenAI just sunk half of the Voice AI startups, and either nobody noticed yet, or everyone in Telco is in shock. 🤔 When OpenAI announces new features, my LinkedIn usually overflows with excitement. Not so much this time, although the impact of changes for companies deploying phone-connected are HUGE! 💡 Not only did OpenAI release their realtime api from Beta to GA. They let a new speech-to-speech model off their leashes. 💥 But most importantly, they also released a native SIP interface to their realtime API! There are quite some startups, who only provide an interface to the public phone network for existing speech-to-speech AI services, often without much telco moat, but relying mostly on Twilio and others. And they are in hot water now. Why? Because everyone can now just grab a phone number from Twilio, forward the number to the SIP URI provided by OpenAI, throw in some prompts and feed it with their data, and off they go. Ok ok, if your startup provides proper tool calling to perform more sophisticated integration tasks, you're still good, because that takes some effort to get right. But the voice interface for AI assistants just became commodity. Anyways, time to step up your game as AI Voice Assistant company. Use-cases will now become more complex in order to differentiate. 🆘 To assure their service quality continuously and automatically, we're here to help with Sipfront! Don't hesitate to reach out to me! But what really excites me is that OpenAI chose to use the open source sipgo library, which Sipfront is proud to regularly sponsor with paid feature requests. Secure and encrypted SRTP calls coming to the library powering one of the most popular AI companies? Sure enough, stay tuned!

334

57 kommentarer

🌐 Exciting momentum for 6G coordination! It was a privilege to review the outcomes of the recent 3GPP & O-RAN ALLIANCE Joint Workshop on 6G, held in Sophia Antipolis, France (April 24–25, 2025). With over 400 global participants and 23 key contributions, the workshop aimed to align visions, clarify roles, and foster synergy between 3GPP and O-RAN. 🔑 Key highlights: 3GPP defines the 6G specifications and architecture, while O-RAN complements these with specifications that enable real-world deployment and automation. Strong focus on AI/ML, fronthaul, SMO architecture, and cloud-native infrastructure. Shared commitment to reduce fragmentation and duplication across organizations. 🤝 Collaboration is key — and not just between 3GPP and O-RAN. As we move toward 6G, it’s crucial to broaden the ecosystem and also engage with The Linux Foundation and its open-source communities, to support rapid, secure, and interoperable innovation across software, infrastructure, and testing. Let’s continue working across industry boundaries to co-create the next generation of mobile networks — open, intelligent, and cloud-AI-native by design powered by 🐝 eBPF. 🌍📶 #6G #3GPP #ORAN #LinuxFoundation #OpenNetworking #AI #CloudNative #TelecomInnovation #OpenRAN #MobileNetworks #FutureOfConnectivity #SMO #AIinTelecom

99

7 kommentarer

🚀 AI-Native RAN: Benchmarking the Future of Intelligent Networks As we move towards 6G, the integration of AI into the Radio Access Network (RAN) is no longer optional—it is foundational. But one challenge persists: how do we objectively measure AI-nativeness in RAN? That’s why we propose the AI-RAN Benchmarking Framework—a structured, transparent methodology to evaluate AI integration across four stages: 1️⃣ AI as Add-on – isolated optimization of specific RAN functions 2️⃣ Native AI Support – AI-ready architectures with full data lifecycle management 3️⃣ AI as a Service (AIaaS) – offering AI workloads and LLM-driven optimization as monetizable services 4️⃣ Intent Orchestration – LLMs interpreting business intents and translating them into autonomous network actions 🌐 This framework provides objective, reproducible, and real-world relevant benchmarks—empowering operators, vendors, and researchers to compare solutions fairly and accelerate innovation. The journey to AI-Native RAN is not just about adding AI—it’s about transforming the RAN into an intelligent, multi-purpose platform that can adapt, optimize, and evolve in real time. 👉 We invite the global ecosystem—operators, vendors, startups, and researchers—to join us in refining and adopting this benchmarking framework. Together, we can drive the evolution of truly AI-native 6G networks. AI-RAN Alliance https://ai-ran.org #AINative #RAN #6G #AI #AIonRAN #TelecomInnovation #AINativeRAN #FutureNetworks #LLM #AIaaS #IntentBasedNetworking

69

The #JioBrain powered Neural Ops AI platform redefines how we operate 5G networks at scale. The Neural Ops platform model is trained on #5G network function data - may it be alarms, clear codes, log signatures, or other metrics such as real-time interface traffic patterns - either on the 5G #SBA interfaces, or even IP/MPLS links. This data is continuously collected and fine-tuned on the model as the network operates. Basis this model, #AI #inference takes place during operations in a #decentralised manner, and the platform is able to forewarn impending negative network events with a root cause generated automatically. This pin points issues even before they happen, with a technical root cause and remdial action using AI, and a correction workflow is also triggered as required. There are many other useful insights shown by the platform - such as realtime data correlation across layers (#IP/MPLS, #OS, #Cloud fabric, #Containers & 5G NFs), cluster by cluster analysis of traffic - all done through #AI With this platform, the manual effort required to manage and operate the network is eliminated to a large extent. This platform is operationalised at “#Jio scale” in the True #5GSA Network. The Neural Ops platform will be a default in the Jio #6G core network, as the standards evolve to a combo #4G/#5G/#6G core.

123