STL Partners：2026代理型AI：实现运营商数据的AI就绪研究报告（英文原版+译版）（22页）.pdf_三个皮匠报告

Network Innovation AGENTIC AI: MAKINGOPERATOR DATA AI READY Todeliver on their AI visions(especially agentic AI),operatorsmusttransform their datafoundations andcapabilities to avoid “garbage in,garbage out”.Weidentify the steps operators can take to get their dataready for AI on the journey towardsLevel4+ autonomy. Kuba Smolorz, Senior Consultant|kuba.smolorz@stlpartners.comDarius Lloyd, Consulting Director|darius.lloyd@stlpartners.com|January 2026 Foreword Methodology This report presents insights from an ongoing research programme exploringoperator’snetworkautomationand AItransformationjourneys. This reportbuildson STL Partners existing expertise inthe space through in-depth interviews with8 telecoms operatorsand5 technology vendors in theautomation, AIand data spaces. The aimof the research isto understand how AI,particularly agentic AI, can support operatorambitions towards Level4+ network autonomyand, importantly, the steps operators musttaketogetstarted,makingpractical strides towards their longer term objectives. Editorial independence This report has been prepared by independent consulting and researchcompanySTL Partners andwas commissioned byAmdocs. STL Partners maintainseditorial independence. Mentions or allusions to companies or products inthis document are intended as illustrations of market evolution and are not included as endorsementsor product/service recommendations. Executive Summary Operators globally are pushing towards their ambition ofLevel4 network autonomy by the end of thedecade.A key component of this drive will be the successful implementation and integration ofagenticartificial intelligence (AI)capabilities into their automation practices, unlocking the value oftrusted intelligence–the combination of intelligent AI tooling withtransparency, audit trails andtesting/optimisation with human-in-the-loop cover. Trusted intelligence and the broader shift towards AI-native networking is fundamentally underpinnedbyhigh quality data and operator knowledge. Without this, flashy tools will be limited by“garbage in,garbage out”and AI will not truly be able to transform day-to-day network operations. Data refreshes are easier said than done and buildingquality data platforms and systems that are AI-ready will take time and investment. In this report, we outline where telcos should start on the datajourneyto make quick progress and sustain investment with faster ROI, as well as the guiding NorthStar principles telcos must follow to build strong data foundations and the ideal telco networkontology. Our recommendations for telcos on their data journeys At a highlevel, there are three major takeaways from this report: 1.Start data transformation use case by use case:Operators should begin their datatransformation journey to get agenticAI data-ready by focusing on individual use cases first.This will provide a tangible outcome to work towards and, ideally, shorter term ROI to enablereinvestment. It will also make it easier to align budgets as initiatives begin in a single team ordomain. These use cases should tackle complex automation challenges that build on existing capabilities(for example,taking existing fault management workflows and augmenting with agentic-drivenresolution recommendation and autonomous remediation) or tackle a broader use casearea/domain (such asautomatingthecustomer journey from quote to fulfilment). 2.Donot build in siloes:Though operators should start use case by use case, they should ensurethey (and their partners) are consolidating knowledge and technology built from each use casedeployed and understanding how this can be used in future use cases. This will acceleratefuturedeployments of agentic use cases as initiatives scale and balance speed of maturity with thebuilding of technical debt. Ensure transformation teams are building with theNorthStar principles of best practicedataarchitectures, data governance and data technologiesin mind(asdetailedlater in thisreport).This will ensure operators progress on the three Cs of data(consolidation, contextualisation andcorrelation)1, building towards true operator knowledge and an ontology of the network and its services.An initial step and focus for operators on this journey should be the consolidation ofdata systems down from hundreds, or even thousands, towards single-digit numbers. When thinking aboutNorthStar data principles,particularly as operators seek to embed a“datafabric”within the organisation,telcos should consider the following critical success factors: •Acentral team is empowered (withbudgetandauthority) to enforce standards and driveadoption. •Domain teams(for example,within the network)are incentivised (measuredandrewarded) forowning data products (not just consuming),otherwise they may see itas additional burdenrather than value. •Governance model (federated) is clearly defined with domain stewards,across‐domain steeringcommitteeandbudgets for data product enhancements. •The framew