In March, SAP launched visu­al inspec­tion capa­bil­i­ties pow­ered by AI for its Asset Per­for­mance Man­age­ment (APM) plat­form. Man­u­fac­tur­ing, util­i­ties, and oth­er equip­ment-heavy sec­tors typ­i­cal­ly spend sig­nif­i­cant resources on man­u­al inspections.

Dur­ing a recent ASUG web­cast, Simon Lee, Prod­uct Mar­ket­ing Man­ag­er for APM, walked view­ers through the prac­ti­cal appli­ca­tions of these new inspec­tion tools for orga­ni­za­tions. ​“We aim to help you stream­line inspec­tion process­es, reduce over-main­te­nance, and increase reli­a­bil­i­ty. But the dri­ver here is to help reduce the costs of exe­cut­ing inspec­tion,” said Lee.

Here are sev­en key insights from the webcast: 

1. Uni­fied Data: No More Dis­crep­an­cies Between SAP APM and S/4HANA

SAP engi­neered a ​‘one domain mod­el’ — ensur­ing both S/4HANA and APM access and update iden­ti­cal mas­ter data records. Every ele­ment remains synced across plat­forms, from equip­ment records to work orders and mea­sure­ment points to main­te­nance noti­fi­ca­tions, elim­i­nat­ing data con­sis­ten­cy prob­lems. Lee said the seam­less con­nec­tion ​“cre­ates a closed-loop process between your reli­a­bil­i­ty and main­te­nance man­age­ment organizations.”

2. AI Trans­forms Labor-Heavy Visu­al Checks

For one com­pa­ny, over 50% of indus­tri­al pre­ven­tive main­te­nance involves inspect­ing equip­ment visu­al­ly. Asset-heavy com­pa­nies must send spe­cial­ists to per­form these assess­ments and fre­quent­ly halt oper­a­tions while inspec­tions occur. APM now inte­grates cam­era and drone imagery with ana­lyt­i­cal tools that part­ly auto­mate inspec­tion process­es. Sev­er­al sec­tors, par­tic­u­lar­ly rail, util­i­ties, min­ing, and man­u­fac­tur­ing, stand to ben­e­fit most since their facil­i­ties demand reg­u­lar visu­al assessment.

3. Visu­al Inspec­tion Inte­gra­tion with Exist­ing Mon­i­tor­ing Systems

Reli­a­bil­i­ty engi­neers can assign cam­era-based mea­sure­ments to their crit­i­cal equip­ment records along­side tra­di­tion­al sen­sor read­ings. Cus­tom thresh­olds and mon­i­tor­ing rules deter­mine when images indi­cate poten­tial issues. When prob­lems appear, the sys­tem cre­ates alerts lead­ing to main­te­nance noti­fi­ca­tions in SAP S/4HANA. The plat­form even sup­ports com­plex con­di­tions by com­bin­ing visu­al data with oth­er mea­sure­ments. For main­te­nance teams, this cre­ates a sin­gle dash­board show­ing com­plete asset health sta­tus regard­less of data source.

4. Lever­ag­ing the SAP Part­ner Ecosys­tem to Devel­op Visu­al AI Models 

Rather than devel­op­ing its own pro­pri­etary visu­al AI mod­els, SAP cre­at­ed the infra­struc­ture and APIs nec­es­sary to inte­grate with part­ner-built or cus­tomer-devel­oped AI solutions. 

Lee clar­i­fied: ​“SAP is not bring­ing stan­dard visu­al-based AI mod­els for visu­al inspec­tion at this moment. We are pro­vid­ing the means for part­ner solu­tions, their visu­al AI mod­els, or your own AI mod­els to do the AI-based deriva­tion or quan­tifi­ca­tion of the indicators.” 

The ecosys­tem strat­e­gy allows cus­tomers to lever­age exist­ing AI invest­ments or select spe­cial­ized solu­tions from SAP’s part­ner network.

5. Swiss Fed­er­al Rail­ways Demon­strates Prac­ti­cal Appli­ca­tion and Value

Dur­ing the web­cast, Lee detailed how SAP worked with Swiss Fed­er­al Rail­ways (SBB) lever­aged imple­men­ta­tion test­ing. SBB now mon­i­tors train pan­tographs while they oper­ate using track­side cam­eras. Man­u­al inspec­tions pre­vi­ous­ly required remov­ing trains from ser­vice. The cur­rent approach cap­tures images dur­ing nor­mal oper­a­tions, with AI mea­sur­ing com­po­nent thick­ness and gen­er­at­ing main­te­nance noti­fi­ca­tions based on actu­al con­di­tions rather than pre­set sched­ules. This approach has reduced main­te­nance expens­es for SBB while enhanc­ing over­all sys­tem performance.

6. Build­ing AI Feed­back Loops

When tech­ni­cians review results, they mark whether AI analy­sis got it right or wrong. Each cor­rec­tion teach­es the sys­tem, while each con­fir­ma­tion strength­ens its pat­tern recog­ni­tion capa­bil­i­ties. Lee called this approach ​“feed­back-based learn­ing” or ​“rein­force­ment learn­ing,” not­ing that this mech­a­nism ​“enables you to con­tin­u­ous­ly improve the AI mod­el.” Keep­ing per­son­nel involved builds trust in the sys­tem and pro­gres­sive­ly enhances its accu­ra­cy over time.

7. Side-by-Side Inte­gra­tion Capabilities 

The side-by-side inte­gra­tion archi­tec­ture allows cus­tomers to use their own AI ser­vices, part­ner solu­tions, or sys­tem inte­gra­tor imple­men­ta­tions while con­nect­ing to APM through stan­dard­ized data inte­gra­tion capa­bil­i­ties, APIs, and data ser­vices. SAP pub­lish­es all nec­es­sary tech­ni­cal guides for these con­nec­tions on its help por­tal and API hub.

For asset-inten­sive orga­ni­za­tions, the ben­e­fits can already be seen in ear­ly adopters, with inspec­tion cost reduc­tions exceed­ing 50% and many no longer rely­ing on exter­nal AI con­sul­tants. By accom­mo­dat­ing exist­ing AI invest­ments and part­ner tech­nolo­gies, SAP makes imple­men­ta­tion more straight­for­ward, help­ing main­te­nance teams real­ize val­ue faster.

Watch the full web­cast replay here.