2024 Catalyst Projects

See Innovation Come To Life

At the heart of innovation at Innovate Asia, 10+ Catalyst projects will debut their groundbreaking innovations live in the expo hall and on the Innovate stage.

Harnessing the collaborative global force of the greatest industry minds from global organizations, our Catalyst project teams are pioneering solutions to propel industry innovation and growth through Open APIs, ODA, AI, and automation.

Experience first-hand their inventive and trailblazing demonstrations. Delve into the challenges tackled, use cases explored, and solutions forged. Connect with these visionaries to discover how you can leverage their achievements to align with your business objectives and advance future outcomes.

Catalyst Champions include:

Browse Catalyst Projects

OmniBOSS - Phase II

OmniBOSS Phase II proves that even with minimal effort, agentic AI can provide contextual insights from real operational data, closing the gap between practice and execution while preserving telecom expertise for the future. OmniBOSS is an Agentic AI platform that revolutionizes how Communication Service Providers (CSPs) operate their B/OSS environments by embedding domain knowledge, best practices, and AI-driven oversight directly into operational workflows. Unlike traditional systems that passively store configurations and metrics, OmniBOSS proactively monitors, evaluates, and recommends corrective actions across B/OSS layers — acting as a real-time expert assistant. In Phase I, OmniBOSS demonstrated a working prototype of Agentic AI for B/OSS best practices using simulated data. The goal was to prove the conceptual feasibility: AI agents can understand, enforce, and recommend operational best practices across TM Forum-aligned domains like alarms, thresholds, and inventory. Phase II builds on this foundation by extending the solution in two key ways: 1. Real-World Data Validation We evolve from simulation to validation against real-world data samples (anonymized or exported from live systems). This elevates credibility by showing how agents respond to actual operational complexity, not just theoretical cases. 2. New Asset – Best Practice Coverage Heatmap We introduce a visual analytics layer that displays which TMF API areas are fully, partially, or not yet covered by best practice enforcement. This new asset acts as a strategic roadmap for CSPs to prioritize improvements and track operational maturity.

URN: C25.5.888

Project details

Multi-layer cross-domain multi-agent system: Building a "dark-factory" processing mode for network faults management

Communication Service Providers (CSPs) and vertical industries are increasingly demanding automated, intelligent fault management in transport networks. Today, fragmented systems and poor OSS-OMC integration result in manual-heavy, inefficient fault handling processes. As one top CSP leader shared: "We urgently need end-to-end automation in fault management to achieve true service continuity. Manual intervention not only slows down resolution but also increases operational risks." The current challenges — including inaccurate fault identification due to rule-based limitations, slow diagnosis due to reliance on manual tools, and delayed repair due to inefficient communication — severely impact service quality and operational costs. Solving these issues through a multi-layer, cross-domain agent architecture powered by large language models and digital twin technologies will enable accurate fault recognition, rapid root-cause analysis, and secure, automated resolution. This transformation will drastically reduce MTTR (Mean Time to Repair), improve network resilience, and lower dependency on individual expertise, enabling more scalable and intelligent operations. Project Focus: Solving transport network fault handling issues. Traditional methods rely on manual/fixed-rule alarm identification, causing missed/redundant work orders. Cross-domain systems lack coordination; OSS and OMCs don’t interact effectively, hindering automated closed loops, with inefficient manual intervention. Diagnosis takes over 30 minutes (hours for complex faults) via manual/standalone tools, limited by experience transfer. Repairs involve frequent communication or manual operations, with low safety/efficiency. The project uses LLM, Agent tech and small-model algorithms to build a multi-layer cross-domain Multi-Agent system covering network devices. It enables end-to-end automated fault handling via Agent interaction and digital twin simulation. Zhejiang Mobile’s pilot showed 100% alarm coverage, reduced redundant WOs, faster diagnosis, lower costs, and enhanced efficiency. O&M Copilot cuts resolution to 40 minutes, minimizing outages, boosting stability/experience, and driving intelligent, efficient transport network operations.

URN: C25.5.859

Project details

PRISM-AI

Modern telecom networks face persistent challenges in fault detection and resolution due to fragmented operational data and loosely coupled system dependencies across heterogeneous domains. Traditional monitoring systems, reliant on inventory data traversing multiple intermediaries, suffer from latency, inconsistency, and contextual loss—hindering effective impact analysis. Challenges with heterogenous domain: * Fragmented Operational Data: Data related to devices, services, and customers is scattered across multiple systems, making it hard to maintain consistency and context. * Limited Cross-Domain Visibility: Faults that span across network domains are difficult to analyze due to loosely coupled telemetry and topology awareness. * Ineffective Impact Analysis: Without unified and real-time context, assessing the impact of faults becomes unreliable and slow. * Manual Resolution Bottlenecks: Lack of automation and intelligent correlation leads to longer mean time to repair (MTTR) and reduced service reliability. This Catalyst focuses on a transformative Agentic AI framework for intelligent fault correlation and autonomous resolution. Central to the solution are the A2A & Model Context Protocol, which enable dynamic synchronization of network inventory and topology across agents. This ensures real-time, context-rich fault and impact analysis and significantly enhances precision and responsiveness. The architecture adheres to the TM Forum Open Digital Architecture (ODA) and integrates standardized TMF Open APIs along withTMF921, TMF931 for seamless orchestration, telemetry ingestion, and incident lifecycle management. By leveraging AI to correlate cross-domain events and automate remediation, the solution aims to reduce Mean Time to Repair (MTTR) by ~ 25%-30% and elevates service reliability. A key innovation lies in consumer-facing capability wherein through Camara APIs, end-users can access real-time network insights, opening new avenues for transparency, trust, and monetization. "Our network spans multiple domains - Optical, IP, Power - each with its own tools and data silos. When issues arise, faults often cross these boundaries, making root cause isolation inefficient. By investing in cross-domain, AI-driven service assurance, we can cut false incidents and reduce mean time to repair by up to 50%, directly improving customer experience, and helping to improve Autonomous Network rating" – Utsav Jain, Senior Manager – Network Monitoring at BT The heterogenous nature of modern Fixed Network architectures has the consequence of making Service Assurance reactive, with lots of false positives and reduced Quality of Experience (QoE). Further complications result from siloed and static rule-based fault detections that lack service context. Finally, the fault resolution journey is based on decisions taken manually by network operators following sometimes outdated runbooks. The solution to the above problems lies in a holistic approach that can understand and leverage User Intent and provides AI-driven Unified Assurance with Multi-Domain Fault Correlation, real-time Network Insights, and Autonomous Resolution.

URN: C25.5.875

Project details

AI-driven proactive Customer-Centric O&M: Empowered by multi-agent and digital twin

Telecom operators are racing to advance toward AN4-level autonomous networks, yet they face severe challenges: delayed user perception assurance, where passive manual responses fail to predict issues in advance, and slow complaint handling continues to erode customer satisfaction; imprecise fault analysis, making it difficult to reasonably prioritize critical issues amid redundant work orders; and inefficient fault handling, with low accuracy in root cause identification, siloed systems, and unvalidated solutions. These not only prolong the Mean Time to Repair (MTTR) but also threaten network stability. Addressing these challenges will deliver significant business value: drastically reducing operational costs by minimizing manual operations, enhancing network reliability through faster and more accurate fault resolution, and strengthening competitive advantages by upgrading service quality. For users, this means fewer network disruptions, quicker problem resolution, and consistently superior experiences—turning dissatisfaction into trust. This project integrates signaling analysis large models, spatiotemporal analysis large models, multi-agent collaboration, and digital twin technology, shifting the focus of operations from "network-centric" to "customer and business-centric." It enables proactive issue prevention, automated end-to-end cross-domain process closure, and risk-controllable solutions, reshaping the operation model to bring "ultimate user perception" to both operators and customers. The core value of operators' transition to L4 autonomous networks lies in achieving proactive and automated operations. However, the current passive, incident-driven O&M model keeps customer complaints high, with three key issues: 1. Lagging user perception assurance: Inadequate optimization of service quality improvement processes and weak ability to locate quality issues result in reactive operations that fail to "identify problems before users". Meanwhile, manual, lengthy complaint handling with bottlenecks leads to inefficiency and reduced customer satisfaction. 2. Inaccurate fault analysis: Massive alarms and work orders lack metrics for assessing impacts on services and user perception, treating all equally. The phenomenon of "one fault generating multiple orders" also exists, failing to prioritize critical fault handling. 3. Inefficient fault disposal: Low accuracy in root cause identification, over-reliance on expert experience for solutions, lack of automatic collaboration between cross-domain systems, and absence of simulation verification not only waste time and effort but also cause misjudgments. This leads to long Mean Time to Repair (MTTR), uncontrollable network operation risks, and impacts on network quality and customer service guarantee. These issues not only increase O&M costs but also directly affect customer retention. For instance, the churn rate of high-value users has risen year-on-year due to undetected service quality issues. Addressing these challenges will reshape the competitiveness of Communication Service Providers (CSPs): Signaling and spatiotemporal analysis models can enhance problem prediction accuracy; multi-agent collaboration enables "minute-level" closed-loop handling of cross-domain faults; digital twin verification reduces operational risks. For operators, users' demand for a "seamless network experience"—consistently stable and smooth service—has become core. Traditional "firefighting" O&M not only consumes resources but also erodes user trust, a fatal flaw in the digital era. For vertical industries like finance and healthcare, a more stable network will accelerate the implementation of their digital services, ultimately achieving value co-creation between CSPs and industry clients.

URN: C25.5.868

Project details

Intelligent Multi-Domain Cross-Disciplinary Network O&M Capability

Intelligent Multi-Domain Cross-Disciplinary Network O&M Capability solves critical network operation failures caused by human configuration errors and cross-domain silos. Current fragmented systems result in >70% service outages from manual mistakes in multi-vendor environments, delaying fault resolution for hours. Our solution integrates four patented modules: Wing Script: Pre-change script audit via conflict detection prevents erroneous configurations. Large AI configuration models combine with small models applied in IP resource adjustment system for automated IP network vulnerability identification. Wing Simulation: Protocol behavior simulation using routing/flow inputs predicts routing/forwarding tables. Cross-vendor (Huawei) heterogeneous simulation enables full-network coverage. Wing Topology: Automatically builds real-time updated network physical topology / dynamic network service routing flow topology, enables inspection and maintenance capabilities based on service flows, predictive maintenance, and circumvents large-scale failures.Integrating end-to-end ping/trace, log analysis, and alarm correlation. Wing AI-Config:Pulls approved plans & scripts, auto-executes deployments, alerts on errors. Validates scripts against plans, restricts high-risk commands, audits execution for compliance & smarter change control. Business Impact: • Zero mass service disruptions • 80% faster MTTR • 40% OPEX reduction Innovation: First integrated AI agent merging pre-audit, multi-vendor simulation, real-time digital twin (99% accuracy), and self-healing automation – transforming siloed operations into error-proof networks.

URN: C25.5.890

Project details

Autonomy accelerated: Intent to impact - Phase II

Our Catalyst addresses one of telecom’s most ambitious challenges: achieving fully autonomous networks at scale; networks that self-configure, self-heal, self-optimize, and can sense, think, and act. As Phase II of the Moonshot Catalyst, Evolving to Full Network Autonomy, this project advances the journey toward Level 4+ autonomy. As communications service providers (CSPs) operate across increasingly complex, multi-technology environments, manual operations and reactive monitoring are no longer sustainable. These limitations lead to degraded customer experience, higher operational costs, and growing risk. To meet rising B2B expectations for seamless, always-on connectivity, CSPs must transform their operating models, unlocking new business value through autonomy, intelligence, and efficiency. Our Catalyst accelerates this vision by demonstrating Level 4+ Autonomous Network capabilities in a high-impact B2B context. It integrates TM Forum’s intent-based architecture (IDAN) and standardized Intent APIs with a digital twin-powered closed-loop system, enabling proactive assurance, predictive action, and intelligent automation. Using AI and Agentic AI, we boost sales efficiency, reduce operational risk, enhance customer satisfaction, and shorten resolution times. What sets our approach apart is its practical application of advanced TM Forum assets to deliver measurable business outcomes. Fully autonomous operations are not just possible, they are essential, scalable, and transformative for the industry.

URN: C25.5.861

Project details

AI-Driven customer experience network investments

Mobile operators still decide network investments using sparse drive-tests, siloed OSS data, and slow manual analysis. The result is delayed fixes, overspending on CAPEX where it doesn’t lift customer experience, and OPEX tied up in fieldwork. Our Catalyst introduces a DePIN-powered, AI- driven loop that turns customer devices into a privacy-preserving sensing network and converts insights into immediate action. With explicit consent, users contribute quality-of-experience (QoE) and radio metrics via a lightweight app and receive on-chain token rewards. This creates a high-resolution, low-cost data fabric with transparent incentives and verifiable provenance. We then use the crowdsourced data with OSS/KPI, probes, and site parameters to build a network digital twin. ML models perform root-cause analysis, predict congestion weeks ahead, simulate “what-if” scenarios, and recommend zero-touch parameter changes or targeted build-outs ranked by CX gain per dollar. Operators cut drive-test costs, accelerate time-to-detect/repair, and direct CAPEX to the neighborhoods and venues that matter most—boosting NPS and revenue while reducing OPEX. End users see faster, more reliable service precisely where they live, commute, and attend events.

URN: C25.5.901

Project details

Voice-only location and ride-hailing services for seniors and travelers 'TeleNavi'

With internet evolving at a breakneck pace, the elderly are becoming increasingly unfamiliar with mobile phones. Many of them are unable to use E-maps, GPS navigation, and ride-hailing apps, even smartphones, which puts them at a competitive disadvantage of being difficult to hail a cab on the roadside for a long time. The “Digital Travel Dilemma” traps the lifestyle and restricts the quality of life and the consumption level of the elderly. Due to the difficulty in locating users and the high costs and low efficiency of human customer service or audio order listening by taxi drivers, the current use and promotion of ride-hailing service hot line are very challenging. TeleNavi is the world’s first "Voice-Only Location and Navigation Tech and Service ", designed for elderly users, international travelers, and those in low-network-coverage areas, delivering seamless travel assistance through voice positioning. Leveraging TMF Open API, GSMA Open Gateway API and AI LLM, TeleNavi enables voice-only positioning and ride-hailing services via phone calls, making travel more convenient and inclusive. No apps, no GPS, no 4/5G mobile data, no smart devices and no complicated steps, just a basic feature phone to make a simple phone call can solve it all.

URN: C25.5.898

Project details

AI and digital twin-driven customer experience optimization

In the rapidly evolving telecom landscape, traditional customer satisfaction management methods struggle to keep pace with the dynamic demands of modern users. This project introduces a data-driven customer experience ecosystem that leverages AI and digital twin technologies to transform how CSPs understand, predict, and enhance user satisfaction. By integrating real-time network performance, service delivery, and behavioral analytics, the solution enables CSPs to proactively address issues, optimize resource allocation, and align customer needs with business outcomes. Unlike conventional approaches, this framework shifts from reactive feedback loops to predictive experience design, ensuring that every interaction contributes to measurable business value. CSPs face mounting pressure to deliver seamless, personalized experiences in an era where user expectations are shaped by hyper-connected digital ecosystems. Traditional methods of measuring satisfaction—such as surveys and static KPIs—are limited by: Fragmented data: Network metrics, service feedback, and user behavior remain siloed, preventing holistic root-cause analysis. Delayed responses: Reactive decision-making based on outdated feedback hinders real-time issue resolution (e.g., network latency or service gaps). Misaligned priorities: Satisfaction improvements often lack clear connections to revenue growth, making ROI justification challenging. Solving these challenges will redefine how CSPs operationalize customer experience. As Başar Günyel, Senior Manager of Network Quality at Vodafone Türkiye, states: "Our customers’ happiness and experience are our main-focus areas, and we measure this through complaints, surveys, and other feedback mechanisms. NPS surveys provide valuable insights into customer sentiment, and we leverage various network data sources to correlate this feedback with actual network performance.“ This approach to customer satisfaction management will foster a more personalized, responsive, and efficient customer experience. Ultimately, by enhancing network performance and reliability, it empowers communications service providers to proactively meet evolving customer expectations, ensuring sustained competitiveness and market leadership.

URN: C25.5.886

Project details

AI marketing brain in telecoms

In an era where over 70% of CSPs struggle with fragmented data and static customer profiles, the AI Marketing Brain introduces a transformative solution. By leveraging a Large User Model (LUM) built on Graph Neural Networks (GNNs) and Transformer-based AI, this Catalyst enables real-time, hyper-personalized marketing strategies. The system integrates multi-domain data (e.g., CRM, billing, network usage) to predict user behavior, automate campaign execution, and optimize pricing dynamically. Unlike traditional marketing, which relies on manual workflows and delayed analytics, the AI Marketing Brain reduces decision latency to seconds, driving measurable outcomes like 18% ARPU growth in 5G upselling trials. This project aligns with TM Forum’s DT4DI standards, ensuring scalability across 2B/2C scenarios while addressing privacy compliance through synthetic data training. CSPs face a critical challenge: declining customer retention rates and stagnant revenue growth amid rising operational costs. For example, Telkomsel’s 5G adoption campaigns previously failed due to generic targeting and delayed feedback loops. As one executive noted, “Our marketing teams were working with last year’s data, while user behavior evolved in weeks. We needed a system that could predict churn and act before it happened.” The AI Marketing Brain directly addresses this by enabling proactive, data-driven decisions. For B2C scenarios, it reduces manual effort by 40% while increasing campaign ROI. For B2B, it streamlines FMC (Fixed-Mobile Convergence) scenarios by integrating enterprise usage patterns with consumer behavior. Solving these pain points will allow CSPs to shift from reactive cost centers to agile, revenue-generating departments. The solution seamlessly integrates operational data, business domain insights, and third-party sources to construct a comprehensive and granular customer profile. Leveraging advanced artificial intelligence models—including graph analytics, machine learning, and real-time event detection—it automates decision-making and enables large-scale personalized customer engagement. For instance, by analyzing behavioral data and usage patterns, the system can precisely identify a segment of 4G users in the courier industry and significantly enhance 5G adoption through targeted voice plan offers. Additionally, it drives higher conversion rates via automated, graph-based revenue optimization strategies.

URN: C25.5.887

Project details

Agentic Intelligence Exchange - A Federated AI Marketplace for Monetizing Telco Intelligence in Real Time

Telcos and enterprises generate vast real-time behavioral signals—quota depletion, device switches, roaming events, usage anomalies but lack the intelligence layer to act on them with speed, context, and precision. Existing engagement models are slow, rules-based, and siloed. Campaigns miss critical moments, network APIs remain under-monetized, and AI systems stay isolated inside narrow use cases. Agentic Intelligence Exchange introduces a federated AI marketplace where telcos, enterprises, and partners co-develop and deploy intelligent agents trained on local data, triggered by real-time events, and orchestrated through standardized Open APIs. These agents deliver hyper-personalized actions across customer channels such as WhatsApp, SMS, Email, RCS without breaching data privacy or operational boundaries. Built on TM Forum’s Open Digital Architecture, the platform enables agents to reason, respond, and evolve in near real time, unlocking measurable outcomes: improved retention, revenue uplift, lower cost-to-serve, and new API-based monetization models for telcos. By shifting engagement from static workflows to agentic orchestration, this Catalyst empowers CSPs and B2B2X partners to commercialize telco intelligence, scale customer personalization safely, and define a new paradigm for AI-native, real-time customer experience delivery. Across the telecom industry, operators and enterprises are flooded with real-time customer signals such as data usage spikes, roaming activations, device changes, recharge behavior, and more. These moments represent real opportunities to engage, retain, and upsell. But acting on them in real time remains a major challenge. The cost of inaction is significant: - $10–30 million in annual churn losses are common among mid-to-large CSPs due to missed engagement windows tied to customer behavior. - 40% of campaigns fail to convert, not because the offer is wrong, but because the timing or channel is disconnected from the customer’s context. - CPaaS platforms are under-utilized. While capable of delivering voice, SMS, RCS and Social Media Messenger or Instant Messaging Apps, fewer than 1 in 5 engagements are triggered by real-time customer signals. - Network APIs remain under-monetized, despite holding valuable insights like quota depletion, LBS, roaming status, SIM Swap, Credit Scoring and device fingerprints. Traditional marketing and engagement systems rely on static rules, batch processing, and delayed execution. Meanwhile, customer behavior changes in seconds not days. This disconnect causes valuable moments to be lost and leaves billions in potential revenue on the table. What’s needed is a way to observe behavior as it happens, understand context immediately, and act with relevance automatically. Agentic Intelligence Exchange addresses this by enabling operators, enterprises, and partners to co-create and deploy intelligent agents that listen to real-time data and deliver personalized, privacy-safe actions instantly. These agents connect across APIs and customer channels WhatsApp, email, webchat without breaching data boundaries. By solving this problem, CSPs can reduce churn, increase engagement ROI, monetize APIs, and transform into AI-native platforms that support scalable, intelligent customer experiences across industries.

URN: C25.5.881

Project details

5G User Perception Improvement Based on Intelligent Computing Board

This proposal aims to address the problem of resource allocation homogenization and the inefficient O&M in traditional networks. The proposal innovatively introduces the intelligent computing board to achieve real-time network state awareness and local reasoning. On one hand, it enables service identification at the base stations, providing differentiated resource guarantees for users; On the other hand, the intelligent computing board transforms passive maintenance into proactive prevention by automatically identifying faults and potential risks, thereby enhancing network stability. Moreover, it can automatically locate root causes and generate solutions with the service opening of the intelligent computing board, reducing manual intervention and improving O&M work order processing efficiency. In summary, the intelligent computing board enhances user perception through differentiated resource allocation and efficient O&M, which strengthens user stickiness and in turn elevates operators' competitiveness. Additionally, the capability of differentiated resource allocation enables operators to generate revenue from customized packages, while efficient O&M reduces their operational costs, bringing substantial commercial value to operators. From an innovation perspective, the intelligent computing board deeply integrates AI technology and the base station, and realizes precise scheduling of network resources through the closed-loop process of “perception-decision-execution”, thereby driving the construction of autonomous networks.

URN: C25.5.895

Project details

Displaying 1-12 of 15 results