5G and the Internet of Things (IoT) represent major growth opportunities for communication service providers (CSPs) in the near future. However, the ability to support emerging use cases in these areas requires business support systems (BSS) that can handle complex business situations and optimize outcomes with minimal manual intervention. Artificial intelligence (AI) is the obvious answer, but adding it on to existing BSS here and there is not sufficient. The latest Ericsson Technology Review article argues in favor of architectural changes to traditional BSS that would allow it to become ‘AI native’. One of the key differences between traditional BSS and AI-native BSS is the fact that the latter enables the various applications within the BSS to share business information with each other in an efficient and secure manner – a critical capability in the emerging 5G-IoT world.

1. Enterprise
Strategic
level
Tactical
level
Operational
level
Business model #1 Business model #2 Common enterprise
resources
Resources common
to all business models,
for example:
∙ Party model
∙ Channels
∙ Credit check service
Credit rules
Channel
serviceCustomer
Contract
Sales
Onboarding
Billing
Settlement
Rating
Customer
support
Intent
Intent Intent
Intent
Intent
Intent
Intent Inten
Intent
Intent Inte
Intent
Intent
ERICSSON
TECHNOLOGY
C H A R T I N G T H E F U T U R E O F I N N O V A T I O N | # 0 1 ∙ 2 0 1 9
AI-NATIVE
BUSINESSSUPPORT
SYSTEMS

2. ✱ AI-NATIVE BUSINESS SUPPORT SYSTEMS
2 ERICSSON TECHNOLOGY REVIEW ✱ JANUARY 30, 2019
The growing need to support disruptive services emerging from the Internet
of Things (IoT) and 5G requires a fundamental transformation of business
support systems (BSS). At Ericsson, we believe that the best way to achieve
this is by forging BSS and artificial intelligence (AI) together to create truly
AI-native BSS.
LARS ANGELIN,
JOHAN SILVANDER
Although AI is of obvious benefit in terms of
business optimization, and has been used in
all sorts of businesses for decades, AI and BSS
have neverbeenintegratedintooneefficient
system.
■ ExamplesofareasinwhichAIisalreadyusedin
conjunctionwithBSSsoftwareincludecustomer
retention,chatbots,revenueandcostpredictions,
customeranalysis,customerexperiencemanagement,
customeryieldoptimization,automation,process
reengineering,simulations,qualityimprovements,
andfraudandanomalydetection.
AIcapabilitiesenableimprovedbusiness
decisiondynamicsandbetterdecisionprecision,
resultinginbetterbusinessperformanceandagility.
Virtuallyallbusinessactivitiescanandwillbenefit
fromAI,andas5GandtheIoTcontinuetoexpand,
thenumberofusecaseswillonlycontinuetogrow.
Thechallengewefaceatpresentisthatthelearning,
insight-buildingandreasoningcapabilitiesofAIin
today’stelcoBSSarenotasstrongastheyneedtobe
tocopewithemergingusecases.
Forthemostpart,AIcapabilitiestodayare
simplyboltedontotelcoBSSonebyone.Butthis
isinefficientintermsoflife-cyclecosts,because
theBSSmustberepeatedlyupgradedtobenefit
fromtheAIalgorithms.Further,astheyareseparate
systems,theBSSinformationmustbetransformedto
fitAIsystems,andviceversa.Amuchmore efficient
alternativeisAI-nativeBSS–thatis,BSSwith
intrinsicAIcapabilitieswheretheAIlogic
BSSandartificial
intelligence–TIME TO GO NATIVE

3. AI-NATIVE BUSINESS SUPPORT SYSTEMS ✱
JANUARY 30, 2019 ✱ ERICSSON TECHNOLOGY REVIEW 3
isanaturalpartofBSSlogicintermsofbothdesign
andoperation.Thisapproachresultsinasystemthat
canhandlemorecomplexbusinesssituations,
generatingmoreoptimizedbusinessoutcomes.
BSSevolutiondrivers
Themaingrowthopportunitiesforcommunication
serviceproviders(CSPs)withinthenextdecadeare
5GandtheIoT,withanestimatedannualvalueof
approximatelyUSD600billion[1,2]. Tocapitalize
onthisopportunity,CSPsmustbeabletosupporta
marketplacewithanecosystemofmanyactorsthat
havetheirownbusinessmodels,whereeachactor
maybebothsupplierandcustomertootheractors.
Businesssupportcomplexityincreasesdramatically
inthisenvironment.CurrenttelcoBSS,whichcan
onlysupportasingleenterpriseshopwithafew
businessmodels,are simplynotuptothetask.
Surveysshowthataclearmajorityofthe
telecommunicationsindustryactorsexpect
AItohavesignificantbusinessimpactinthe
comingfiveyears,affectingboththetopand
bottomlines.TheyalsoexpectAItobring
asignificantcompetitiveadvantagetothe
enterprisesusingthem,growingproportionally
withAIusage.Analystspredictthatenterprises
willinvestinAIcompetence,AImaturityandin
organizationalAIcapabilities[3,4],despite
thecosts[5].
Thereareessentiallythreemainforcesdriving
thecombinedAI-BSSevolution[1].Firstly,business
agilityisahighlyvaluedBSSpropertysincethe
businessitselfevolvesandnewbusiness
opportunitiesemerge.AIplaysakeyroleinboth
identifyingopportunitiesandinshapingthenew
businessmodelstopursuethem.Secondly,the
maturityofthecommunicationsindustryisdriving
everlowerbusinesstransactioncosts,as
demonstratedbyexistingplatformplayerslike
AmazonandAlibaba.Inlightofthis,AI-supported
processautomationandreengineeringarethetools
Artificial intelligence (AI) depends on
software algorithms. At Ericsson, we use
the term AI in its widest sense, including
several subfields such as machine learning,
representation learning and deep learning.
AI-related areas such as natural-language
processing, automated reasoning,
multiagent systems, symbolic learning,
knowledge representation, intelligent
tutoring systems and high-level computer
vision are also included [8].
ARTIFICIAL INTELLIGENCE
Terms and abbreviations
AI – Artificial Intelligence | BSS – Business Support Systems | CRISP-DM – Cross-Industry Process
for Data Mining | CSP – Communication Service Provider | IoT – Internet of Things

4. ✱ AI-NATIVE BUSINESS SUPPORT SYSTEMS
4 ERICSSON TECHNOLOGY REVIEW ✱ JANUARY 30, 2019
ofchoice.Finally,thecloudprovidesanideal
foundationforcontinuousintroductionofAIand
BSSmarketplacecapabilities.Thisisbecausethe
cloudoffersdeploymentflexibility,elasticscaling
andamicro-servicearchitecturethatenablesamore
fine-grainedseparationofconcernsand
componentizationwithloosecoupling.
KeychallengestoaddingAItoBSS
IntroducingAIintoexistingBSSisnot
straightforward.Somechallenges,suchasdata
acquisition,datapipingandtrainingalgorithms,are
obviousandwell known.Others,however,areless
so.Oneexampleofalessobviouschallengeisthe
factthattheinterpretationoftheAIresultsrequires
businesscompetence;anotheristhatmonetization
requiresbothretrainingwithintheorganizationand
redesigningoftheexistingsetofbusinessrulesand
processesandsystemreengineering[5].
Traditional,non-AI-nativeBSSaredividedinto
componentsilos–suchascustomerrelationship
management,catalogs,billingandorder
management–eachwiththeirowninformation.
ThisarrangementcontradictsAIefficiencyand
dynamicsenablement,whichrequireanopen,pan-
BSSinformationandrulesview.WhenanAI
capabilityisaddedtothisenvironment,itistreated
asanadd-on,requiringbothAIandBSSsystem
competence,informationtransformationandin
manycasespartialsystemre-implementationor
reconfiguration.BSSperformanceissuessuchas
latencyandscalingmayarise.
Manybusinesssituationsaremultifaceted,have
manyrootcausesandmayincludebothgainsand
risks.Inmanycases,amainbusinessintent(also
knownasaKPI)mustbebrokendownintoa
combinationofsubintents.Thesewillbebasedon
manydatasetsandalgorithms,andthenbestitched
togetherbyasuper-algorithmtodeliverthemain
businessintent.Thereisalsoariskoflostbusiness
control,asahigh-levelintentmayaffectmanyofthe
lower-levelbusinessrulesandprocesses.Thiseffect
isconsiderablysmallerwhenanintentisintroduced
atlowerlevels,butinthosecasestherewillbeless
businessgain.Theuniquenessofanintentandits
contextmeansthereislittleopportunityforreuseor
experiencebuilding.
EricssonbelievesthatanewBSSarchitecture
stylethatincorporatesAI-nativeproperties–
includingdata-centric,learningloop,intentand
event-drivenlogic,businessrulehierarchy,and
supportforstrategic,tacticalandoperationallevels–
isamuchmoreefficientwaytointegrateAIwith
BSS.WefullyagreewiththeviewthatfutureBSS
andAIwillbeinseparablylinkedandmustmature
together[4,5].
IntroducingintentstoBSS
Anenterpriseisahierarchicalorline-of-command
structureinwhichbusinessrulesatthetopsteer,
alignandcontroltheactivitiesandbehaviorsfurther
downinthestructure.Businessrulesalsosteerall
behaviorinBSS,inpursuitofthegoalofcreatingand
maintainingasuccessfulbusiness.Thestepbetween
abusinessruleandabusinessintentisverysmall;
justasmallshiftofperspective.
Abusinessruleisstaticandstateswhattodoina
givensituation,whileabusinessintentstatesthe
desiredoroptimaloutcomeofagivensituation–
thatis,interpretingwhatthestakeholder’sinterest
isandtryingtodeliverasclosetoitaspossible.
ABUSINESSINTENT
STATESTHEDESIREDOR
OPTIMALOUTCOMEOF
AGIVENSITUATION

5. AI-NATIVE BUSINESS SUPPORT SYSTEMS ✱
JANUARY 30, 2019 ✱ ERICSSON TECHNOLOGY REVIEW 5
Businessintentscanhandlecomplexanddynamic
situationsandallowforfeedbackandcomparison
ofactualanddesiredoutcomes,enablinglearning
andknowledgebuilding[1].
Higher-levelintentsinBSSareoftenexpressed
asbusinessrulesorKPIs.Intentsarefoundatall
levelsofthebusinesshierarchy,supportingboth
top-andbottom-lineoutcomes.Anintentcan
rangeincomplexityfroman‘atomicintent’
toan‘algorithmofintents’thatcombinesaset
ofsubintents.Theterm‘atomicintent’refers
tothesimplestpossibleintentstructure,suchas
“ourcompanywillrunaprepaidbusinessmodel.”
Notethatanatomicintentonastrategiclevel
islikelytofanoutintoseveralintentsona
lowerlevel.Whileintentscanbeformulated
forbothhumanandmachineconsumption,
theymustbestatedinadeclarativeformat
tofacilitateautomationinBSS.
Figure1illustratesthestructureofthe
businessintenthierarchy,whichisdesignedto
mirrorthebusinessrulehierarchy.Anenterprise
musthaveatleastthreedifferentintentlevels
–strategic,tacticalandoperational[5]–allof
whicharetheresponsibilityoftheBSS.
Insoftwareterms,theselevelsareequivalent
torequirements,designandimplementation,
andexecution.
Figure 1 The business intent hierarchy mirrors the business rule hierarchy
Enterprise
Strategic
level
Tactical
level
Operational
level
Business model #1 Business model #2 Common enterprise
resources
Resources common
to all business models,
for example:
∙ Party model
∙ Channels
∙ Credit check service
Credit rules
Channel
serviceCustomer
Contract
Sales
Onboarding
Billing
Settlement
Rating
Customer
support
Intent
Intent Intent
Intent
Intent
Intent
Intent Intent
Intent
Intent
Intent
Intent Intent
Intent
Intent

6. ✱ AI-NATIVE BUSINESS SUPPORT SYSTEMS
6 ERICSSON TECHNOLOGY REVIEW ✱ JANUARY 30, 2019
Thebusinessstrategylevelownsandformulates
theenterprise’stopintents.Forthesakeofsimplicity,
Figure1breaksdownonlyonebusinessmodeltoall
threelevels,butitisimportanttonotethatan
average-sizeoperatorrunsseveraldifferentbusiness
models.Thebusinesstacticlevelisresponsiblefor
designingandimplementingtheintentsofeachof
thesebusinessmodelsattheoperationallevel.This
isnormallyachievedbybreakingdownthestrategic
intentintosmaller,digestibleparts,suchas
subintentswithassociatedrules,informationand
processes.Thebusinessoperationlevel–thecoreof
traditionalBSS–isresponsibleforexecutingto
delivertheintents.Thislevelrequiresfurther
automationtomeetownershipandbusiness
transactioncostrequirements.Allthreelevels
benefitfromAIsupportbuthavedifferentusage
patternsandcharacteristics,asshowninFigure2.
Thedifferencesmostworthnotingareintermsof
repetitions,context,explorationanddata
characteristics.
TheOODAloop
TheOODAloop[6]wasinitiallydevelopedinthe
1970sasanin-combatdecisiontooloftheU.S.Air
Force.OODAstandsforobserve,orient,decideand
act.ManyoftheOODAloop’sbasicconceptsare
foundintoday’ssoftwareagentsystems.Theversion
Figure 2 AI usage patterns are different at strategic, tactical and operational levels
Strategic level Tactical level Operational level
One-off or few
Human interaction
Learning/reasoning
Yes Yes No
Reasoning Learning/reasoning Execution
Feedback Limited Yes, key element Large volumes
Main constraint Quality Quality Time
Clarity of data use Undetermined Limited to own data Deterministic and limited BSS set
Many, external and
large volumes
Context, sources
and data volumes
Data types and
time series
BSS BSS and inference thereof
LargeData size Large but limited Optimized for the intent
All types Many but limited Few and limited to BSS origin
Recurrence Enough to learn Very many

7. AI-NATIVE BUSINESS SUPPORT SYSTEMS ✱
JANUARY 30, 2019 ✱ ERICSSON TECHNOLOGY REVIEW 7
showninFigure3iscomplementedwithexplicit
intentandlearningcapability.
TheOODAloopideaisquitesimple.First,
observeordetectchanges,events,stimuliorother
thingsthathappeninthecontextofinterest,
includinginternalstates.Observationscanbesingle
orunfoldingevents,andtheycanbesimpleor
complexinstructure.Dependingonthedata,AIis
oftenneededtointerpretobservations.Typical
observationsinBSScouldbetheavailabilityofa
customer’susagerecordorthearrivalofapotential
customertoawebshop.
Theorientationstepconsistsofaggregatingand
analyzingtheobservationsthatformthebasisforthe
decisions.Analyzingtheindividualobservations
andaggregatingthemintoacompletesituation
descriptionrequiresmultilevelAIsupport.The
orientationstepinBSSshouldenrichthe
observationwithcustomerdataasmuchaspossible.
ThisdataenablestheBSStoselectthecorrectrating
andchargingparametersforaparticularcustomer
whentheirusagerecordbecomesavailable,for
example,ortoconcludethatavisitortoawebshopis
lookingforanewphonebutseemstobeprice
sensitive.
Toclearlyseparatetheintentfromthedecisionof
actionthatfulfillstheintent,wehaveaddedintentto
ourmodifiedOODAloop.Weachievedthisby
Figure 3 The OODA loop, modified to enable learning and intents
Orientation
Feed
back
Feed
forward
BSS
internal
actions
BSS
external/interaction
actions
Unfolding
external
circumstances
Unfolding
interaction with
environment
External
events
Feed
back
Feed
back
Feed
back
Feed
back
Feed forward
Feed forward
Feed
forward
Observations
Action
execution
Decision
of
actions
Evaluation
and
learning
Intent

8. ✱ AI-NATIVE BUSINESS SUPPORT SYSTEMS
8 ERICSSON TECHNOLOGY REVIEW ✱ JANUARY 30, 2019
dividingthetraditionalOODAdecisionstepinto
twodistinctprocesssteps:intentanddecision.
IntentsinBSSarestatementsofthedesiredbusiness
outcomeinagivensituation.Inthecaseofinvoicing,
thiswouldmeanensuringthattherates/chargeson
theinvoiceareinaccordancewiththecustomer’s
contract.Inthecaseofaprice-sensitivepotential
customervisitingawebshoptolookforalow-priced
phone,theintentwouldbetoconvincethat
individualtobuyaphoneinthemedium-pricerange
ratherthanchoosingtheleast-expensiveoption.
Thedecisionislimitedbytheinventoryof
availablepossibleactions.Aselectionismadefrom
theavailablearrayofactionsthatbestmatchesthe
intent.Itisalsopossibletoenrichthedecisionwith
simulationstopredicttheactionoutcome.The
actionissimplytheexecutionofthedecision.InBSS,
actionsarecarriedoutbybusinessprocessesand
theyresultinbusinessoutcomes.
ExamplesofBSSdecisions(andresultingactions)
wouldincludethedecisiontoapplythestandard
rate/chargeprocessinthecaseofanewcustomer
usagerecord,orthedecisiontoshowaprice-
sensitivewebshopvisitornotonlylow-priced
phonesbutalsomedium-pricedmodelsthathave
receivedexcellentcustomerratings.
Evaluation,learningandfeedbackareessentialto
buildasystemwithoptimalperformancethatcan
adapttobothbusinessandenterprise-external
changes.Anoptimalsystemusestheprocessof
orientation,intentsanddecisionstocontinuously
compareandevaluatebothbusinessoutcomesand
itsowncapabilities.Adaptionmayrequirenewor
additionalhigher-levelanalysis,algorithmredesign
andalgorithmretraining.Sometimes,itisenoughto
havegoodin-operationslearning,suchasa
continuousalgorithmretrainingcapability.An
exampleoflearninginBSScouldbereachingthe
conclusionthatwhendealingwithprice-conscious
webshopvisitors,betteroutcomescanbeachieved
byshowingamixoflow-priceandmedium-price
phoneswithgoodratings,asopposedtoincluding
theexpensivephonesaswell.
TheOODAloopcanhavevariousdepthsof
reasoning,fromdeterministictodeeplearning–that
is,thesameOODA-loopenginecanbeusedto
observe,decideandselecttheproperactionsforall
eventtypes,regardlessofcomplexity.Itcanalsobe
usedrecursivelytobuildlayeredstructureswith
arbitrarydepth–thatis,itcansupportmultilayered
businessprocessesandinteractions.
ItiscriticalthatthepeopleworkinginAI-enabled
processesareabletounderstandthereasoning
behindAI-generatedresults.Allofthestepsinour
modifiedOODAloopcanbeunderstoodand
executedbybothhumansandmachines–which
makesitpossibletoworktogetherinthemost
efficientwaypossiblebasedontheparticular
circumstancesoftheorganization.
Usecase:reducingmanualhandlingofinvoices
Aninvoicing-relatedusecaseprovidesagood
illustrationofhowAIaddsvaluetoBSS.Inthis
scenario,atelecomoperatornoticesanincreasein
thenumberofinvoicesthatrequiremanual
handling,whichiscostlyforthecompany.The
executiveteaminitiatesastrategicprojecttoaddress
theissue.
Theobjectiveatstrategiclevelistwofold:to
establishfactsontheinvoicesituationandtostatea
futureinvoicestrategy–thatis,tosetanintent–
regardinginvoicehandlingandcost.Thestrategy
teampoolsinformationaboutknownchallengesin
invoicingandgathersexternaldatafor
benchmarking.Withthehelpofclassificationand
statisticalanalysisalgorithms,thefollowing
strategic-levelintentsforinvoicesareestablishedby
thestrategicproject:
❭❭ manual handling of less than 1 percent
of all invoices
❭❭ average handling cost per invoice
of less than USD 1.
EVALUATION,LEARNING
ANDFEEDBACKAREESSENTIAL
TOBUILDASYSTEMWITH
OPTIMALPERFORMANCE

9. AI-NATIVE BUSINESS SUPPORT SYSTEMS ✱
JANUARY 30, 2019 ✱ ERICSSON TECHNOLOGY REVIEW 9
Thisuseoftwodimensionsofintentensuresasound
businessbalance,helpingtoavoidpotentialpitfalls,
suchasthepossibilityofreaching0.0001percentof
manuallyhandledinvoicesatanaveragecostofUSD
100perinvoice.
Oncethestrategicintenthasbeenset,workonthe
tacticallevelcanbegin.Theprimarychallengeatthis
stageistounderstandandclassifyallthereasons
whysomeinvoicesrequiremanualhandlingwhile
othersdon’t,andtoselecttheoptimalAIalgorithms
thatcandeliverresultsinlinewiththestrategic
intents.Thetacticallevelbeginsbydefiningan
efficientsubintentstructureandestimatingeach
subintent’syieldtothestrategicintent,inorderto
selectthemostvaluableones.Then,foreach
subintent,itclassifiespossibleAIalgorithmsto
identifythebestones.Importantconsiderations
include:
❭❭ the information requirement and the complexity
❭❭ the volatility of the constituent knowledge
components in the problem, which in turn
determines whether machine learning is enough
or if it must be combined with machine
reasoning or deep learning to create deep
enough or adaptable algorithms
❭❭ feasibility, effort and automation level in
business operations
❭❭ cost estimations, implementation, operation
and support.
Inthistypeofinvoicingusecase,itmakessenseto
introducecustomizedcommunicationpatternsthat
varyaccordingtocustomercharactertype.
Therefore,partoftheworkatthetacticallevel
involvesdefiningthreedistinctcustomerpersonas
–angry,regularanddocilecomplainers,forexample–
andcustomizeanomalyinvoicemessagesforeachof
them.Thenextstepistotestthesedifferent
messagesonasmallportion(1-3percent)ofthe
customerpopulationtofindtherightmessagefor
eachcustomerpersonatoensurethattheir
complaintsareresolvedwithoutescalationto
manualhandling.Selectingonlyasmallfraction
ofthetotalcustomerpopulationreducesthe
businessrisk.
Findingthenecessaryknowledgecomponentsis
aniterativetaskthatrequiresAIsupportandaccess
torelevantinformation.Thetacticallevelisalso
responsiblefortheAIalgorithmlifecycleincluding
design,implementationandoperationallaunch.
Further,itisresponsibleforstatingthenecessary
changestoBSS,soitcanbothcalculateaccordingto
theAIalgorithmsandautomaticallyexecutethenew
behaviorintheinvoicefunctionality.
Thetactical-levelsubintentsfortheinvoicing
usecaseare:
❭❭ invoice input correctness: higher than 99.999
percent
❭❭ invoice anomaly statistics and predictions at
both group and individual level: anomaly type,
costs, volumes, services and dates
❭❭ customer persona classification into three
levels (angry, regular and docile complainers)
with less than 1 percent error
❭❭ customized message success rate above
90 percent.
Thetactical-levelchangestoBSSinterms
ofnewrules,informationandprocessesare:
❭❭ calculate invoice anomalies and recheck invoice
input if anomaly probability is higher than 15
percent
❭❭ determine customer persona complaint
classification with continuous learning
capability
❭❭ customize the message success rate to achieve
continuous learning capability
❭❭ instruct customer support team to “cut it short”
in cases with no anomaly and with angry
❭❭ calculate and expose subintent and intent
outcome, along with their projections and
variance.
Themethodologyofthetacticallevelissimilarto
thatofAI-supportedCRISP-DM[7].Oncethe
tacticallevelstepshavebeencompleted,therole
oftheoperationallevelissimplytoexecutethe
algorithmsandthenewBSSlogicwithasmuch
automationaspossible.Afterashorttrainingperiod,
theCSPcansortouttheinvoicesthatrequire

10. ✱ AI-NATIVE BUSINESS SUPPORT SYSTEMS
10 ERICSSON TECHNOLOGY REVIEW ✱ JANUARY 30, 2019
manualhandling,identifytherootcauses,classify
customersthatsystematicallycomplain,andselect
themostefficientcustomizedmessage.Theerror
rateandthecostperinvoiceareinitiallyquitehigh
butdecreaserapidlybelowthestrategicallystated
intentastheinvoicecommunicationistuned.
Implications
TheinvoicingusecasemakesitclearthatBSSmust
haveanomnipresentAIabilitytobeabletosupport
strategicandtacticalinvestigations,aswellashaving
theagilityinoperationstochangebehaviorto
accommodatenewalgorithms,rules,information
andprocesses.Theinclusionofbusinessmodels–
thatis,thegroupingofrules,informationand
processestunedtoworktogethertodeliverbusiness
outcomesforspecificbusinesssituations–isalso
criticalintheevolutionofBSS.Theintentstructures
mustmirrorthebusinessmodelstructuresandtheir
lifecycles.
AI-nativeBSSrequireanexpansionofthe
businesslogicelements–rules,information/objects
andprocesses–toincludeintentsandevents.The
businesslogicelements,oftenhiddeninside
applications,mustbeexternalizedtosupportthe
conversionofAIfindingstoautomatedBSS
behavior.Thebusinessinformationmustbe
structuredinanontologyandmadeavailabletoall
businesssupportusersandapplications,AIsystems
included,intoabusinessinformationlake.
AI-nativeBSSmustsupportbothrun-timeand
business-design-time.TraditionalBSSareprimarily
Further reading
❭❭ Ericsson, The dawn of machine intelligence, available at: https://www.ericsson.com/en/news/2017/9/the-
dawn-of-machine-intelligence
❭❭ Ericsson, Zero-touch could herald a new era in service provider customer interaction, available at: https://
www.ericsson.com/en/press-releases/2018/5/ericsson-zero-touch-could-herald-a-new-era-in-service-provider-
customer-interaction
builtwithfewconfigurationoptionsforrun-time,
resultinginlessagility[5].WhileitistruethatAIcan
helpabusinessevolveinrunningBSS(forexample
incontinuousdevelopmentandoperationsmode),
thereisnoavoidingthefactthatthisrequiresaBSS
architecturethatisatleastpartiallynew.
Conclusion
Itiswidelyrecognizedthat5GandtheIoTrepresent
themaingrowthopportunitiesforcommunication
serviceproviders(CSPs)inthecomingdecade.To
supportemergingusecasesintheseareas,CSPs
requirebusinesssupportsystems(BSS)thatcan
handlecomplexbusinesssituationsandoptimize
outcomeswithminimalmanualintervention.
Artificialintelligence(AI)istheobviousanswer,but
introducingitintoexistingBSSisproblematicfora
numberofreasons.Instead,Ericssonrecommends
anarchitecturalchangetotraditionalBSStocreate
AI-nativeBSS.Mostsignificantly,thisevolution
requirestheinclusionofanenterprise’sstrategic,
tacticalandoperationallevelsintheBSS,together
withtheintroductionoftwonewbusinesslogic
elements(intentsandevents).Oneofthekey
differencesbetweentraditionalBSSandAI-native
BSSisthefactthatAI-nativeBSSenablethevarious
applicationswithintheBSStosharebusiness
informationwitheachotherinanefficientand
securemanner–acriticalcapabilityinthe
emerging5G-IoTworld.

11. AI-NATIVE BUSINESS SUPPORT SYSTEMS ✱
JANUARY 30, 2019 ✱ ERICSSON TECHNOLOGY REVIEW 11
References
1. TM Forum, Open Digital Architecture, 2018, available at: https://www.tmforum.org/resources/whitepapers/
open-digital-architecture/
2. Ericsson, Unlocking 5G’s revenue potential: a roadmap for operators (press release), February 26,
2018, available at: https://www.ericsson.com/en/press-releases/2018/2/unlocking-5gs-revenue-potential-a-
roadmap-for-operators
3. Forbes, How Artificial Intelligence Is Revolutionizing Business In 2017, September 10, 2017, Louis
Columbus, available at: https://www.forbes.com/sites/louiscolumbus/2017/09/10/how-artificial-intelligence-
is-revolutionizing-business-in-2017/#58ebeab25463
4. Harvard Business Review, Artificial Intelligence for the Real World, January-February 2018, Thomas H.
Davenport and Rajeev Ronanki, available at: https://hbr.org/2018/01/artificial-intelligence-for-the-real-world
5. McKinsey, Smarter analytics for banks, September 2018, Carlos Fernandez Naviera et al., available at:
https://www.mckinsey.com/industries/financial-services/our-insights/smarter-analytics-for-banks?cid=other-
eml-alt-mip-mck-oth-1810&hlkid=d3be9327efb84eccb44a1d8d391d0d8f&hctky=2669978&hdpid=ddc7fd1c-
3815-4675-ae1a-f53e67d88452
6. OODA loop definition available at: https://en.wikipedia.org/wiki/OODA_loop
7. CRISP-DM definition available at: https://en.wikipedia.org/wiki/Cross-industry_standard_process_for_data_
mining
theauthors
Lars Angelin
◆ is an expert in BSS within
Business Area Digital
Services at Ericsson. He has
more than 30 years of
experience in the areas of
concept development,
architecture and strategies
within the telco and
education industries.
Angelin joined Ericsson in
1996 as a research
engineer, and in 2003 he
moved to a position as
concept developer in the
M2M and OSS/BSS areas.
Since 2006 he has focused
on BSS – specifically
business support, enterprise
architectures and the
software architectures to
implement BSS systems. He
holds an M.Sc. in
engineering physics, a Tech.
Licentiate in tele-traffic
theory from Lund Institute of
Technology in Sweden, and
an honorary Ph.D. from
Blekinge Institute of
Technology in Sweden.
Johan Silvander
◆ is a senior specialist in
information management
who has worked at Ericsson
for more than 20 years.
Over the years, his work has
focused on the areas of OSS
and BSS in a variety of
different roles, including
serving as a member of core
architecture teams, working
as a designer, taking
technical responsibility for
integration and installation
projects, and being a test
leader. He holds a Tech.
Licentiate in computer
science from Blekinge
Institute of Technology,
where he is currently
pursuing a Ph.D.
Theauthorswould
liketothank
JörgNiemöllerfor
hiscontributions
tothisarticle.

12. ISSN 0014-0171
284 23-3325 | Uen
© Ericsson AB 2019
Ericsson
SE-164 83 Stockholm, Sweden
Phone: +46 10 719 0000