SAP announced new gen­er­a­tive AI and data gov­er­nance fea­tures for SAP Data­s­phere and SAP Ana­lyt­ics Cloud, enhanc­ing enter­prise-wide access to data and insights as part of its strat­e­gy to estab­lish a com­pre­hen­sive busi­ness data fab­ric for customers.

The inno­va­tions, unveiled March 6 dur­ing the SAP Data Unleashed vir­tu­al event, are meant to pro­vide cus­tomers with a uni­fied data view and sim­pli­fy their land­scapes while retain­ing the busi­ness con­text and log­ic required to effec­tive­ly lever­age gen­er­a­tive AI. 

Juer­gen Mueller, SAP Chief Tech­nol­o­gy Offi­cer and mem­ber of the Exec­u­tive Board of SAP SE, described the data, plan­ning, and gen­er­a­tive AI capa­bil­i­ties as a ​“quan­tum leap” in the company’s abil­i­ty to help cus­tomers dri­ve intel­li­gent busi­ness trans­for­ma­tions with data, not­ing dur­ing the event that ​“imple­ment­ing a holis­tic data and ana­lyt­ics strat­e­gy” has been a key pri­or­i­ty and chal­lenge for C‑level exec­u­tives, includ­ing CIOs, in the SAP ecosystem.

The announce­ments included:

• SAP Data­s­phere will allow orga­ni­za­tions to bet­ter rep­re­sent their real-world use of data through a knowl­edge graph, pro­vid­ing more com­plete con­text from dis­parate data sources to large lan­guage mod­els (LLMs) while inhibit­ing AI-gen­er­at­ed mod­el hallucinations. 
• SAP HANA Cloud vec­tor capa­bil­i­ties, to be intro­duced in the next release of SAP HANA Cloud, will sup­port gen­er­a­tive AI out­puts with busi­ness con­text and fur­ther empow­er cus­tomers to build intel­li­gent data applications.
• Improved inte­gra­tion between SAP Data­s­phere and SAP Ana­lyt­ics Cloud will estab­lish a sin­gle data man­age­ment sys­tem with advanced ana­lyt­ics, to sup­port cus­tomers’ data prepa­ra­tion, mod­el­ing, and planning.
• Joule, SAP’s gen­er­a­tive-AI copi­lot, will be made avail­able in SAP Ana­lyt­ics Cloud, to auto­mate the cre­ation and devel­op­ment of reports and dashboards.
• A ​“com­pass” func­tion­al­i­ty in SAP Ana­lyt­ics Cloud will enable busi­ness users to run data-dri­ven sim­u­la­tions using a chat inter­face to eval­u­ate pre­dic­tive out­comes and con­tin­u­al­ly adjust con­trol­lable vari­ables, opti­miz­ing cus­tomers’ sce­nario mod­el­ing capabilities.
• SAP will expand its part­ner­ship with Col­li­bra to inte­grate its AI Gov­er­nance plat­form with SAP data assets, enabling data gov­er­nance for non-SAP data assets and more ful­ly sup­port­ing cus­tomers’ het­eroge­nous landscapes.

Weav­ing a Busi­ness Data Fabric

SAP Data­s­phere, launched a year ago, is intend­ed by SAP to serve as a ​“com­pre­hen­sive data ser­vice” for cus­tomers, help­ing them to uni­fy their enter­prise data, break down data silos, and cen­tral­ize data man­age­ment. Built on SAP Busi­ness Tech­nol­o­gy Plat­form and first intro­duced as the cloud-based next gen­er­a­tion of SAP Busi­ness Ware­house, SAP Data­s­phere facil­i­tates data inte­gra­tion, cat­a­loging, seman­tic mod­el­ing, data ware­hous­ing, and vir­tu­al­iz­ing work­loads across both SAP and non-SAP data. 

Core to SAP Data­s­phere is the con­cept of an over­ar­ch­ing ​“busi­ness data fab­ric,” a data man­age­ment archi­tec­ture that estab­lish­es an inte­grat­ed, seman­ti­cal­ly rich data lay­er over cus­tomers’ exist­ing data land­scapes. The archi­tec­ture pro­vides seam­less, scal­able access to any data across an enter­prise with­out dupli­ca­tion, while keep­ing busi­ness con­text and log­ic intact.

With these new fea­tures, SAP is updat­ing its data strat­e­gy for the busi­ness-AI era and acknowl­edg­ing that orga­ni­za­tions can­not effec­tive­ly lever­age gen­er­a­tive-AI inno­va­tions with­out trust and trans­paren­cy sur­round­ing the process of data cap­ture and the qual­i­ty of their data. 

By bring­ing data gov­er­nance, data mod­el­ing, and gen­er­a­tive AI fea­tures to SAP Data­s­phere, the com­pa­ny is mov­ing to pri­or­i­tize data qual­i­ty and access while ensur­ing non-SAP and unstruc­tured data (such as text, images, and audio) can also be effec­tive­ly lever­aged in AI-based scenarios. 

“Cap­tur­ing data to make bet­ter deci­sions is an enter­prise tech­nol­o­gy imper­a­tive that’s become increas­ing­ly crit­i­cal as AI — which relies on qual­i­ty data — rev­o­lu­tion­izes every aspect of busi­ness,” said Mueller, in an offi­cial statement. 

Dur­ing the Data Unleashed event, he posi­tioned SAP Data­s­phere as the solu­tion to solve this chal­lenge and urged cus­tomers to ​“put the right data archi­tec­ture in place that will enable you to man­age big data sets with­out data redun­dan­cy, gain valu­able insights, and make intel­li­gent deci­sions — all at scale, with­out rebuild­ing the seman­tics you already have in your SAP system.” 

SAP Data­s­phere Knowl­edge Graph

SAP Datasphere’s knowl­edge graph will take exist­ing data onboard­ed, trans­formed, and inte­grat­ed in SAP Data­s­phere and auto­mat­i­cal­ly cre­ate an ontol­ogy show­ing rela­tion­ships and con­nec­tions between the data, includ­ing busi­ness con­text from oth­er SAP appli­ca­tions like SAP S/4HANA.

Via an ontol­ogy edi­tor, cus­tomers can cus­tomize the ontol­ogy to fit their spe­cif­ic busi­ness needs; then, exist­ing data with­in SAP Data­s­phere will be com­bined with the ontol­ogy to cre­ate a knowl­edge graph, pre­sent­ing and pro­cess­ing the data ​“as a seman­tic web of rela­tion­ships,” as per SAP.

“Knowl­edge graphs are pow­er­ful tools for cap­tur­ing the rich and com­pli­cat­ed rela­tion­ships that exist across your enter­prise by com­bin­ing them into your busi­ness data fab­ric,” Mueller said dur­ing SAP Data Unleashed. Com­bin­ing busi­ness data with rela­tion­al meta­da­ta, he added, is anoth­er ben­e­fit; with knowl­edge graphs, busi­ness users can under­stand more com­plete­ly how the data in their land­scapes is being used, improv­ing their abil­i­ty to uncov­er hid­den con­nec­tions, insights, and usage pat­terns between applications. 

Addi­tion­al­ly, knowl­edge graphs are equipped to answer com­plex, open-end­ed ques­tions — posed by busi­ness users through Joule — and can pro­vide con­text to LLMs, reduc­ing AI-gen­er­at­ed ​“hal­lu­ci­na­tions” in data. ​“Knowl­edge graphs are cru­cial for busi­ness AI,” said Mueller. ​“Lat­est research shows that knowl­edge graphs are more accu­rate than vec­tor engines when deal­ing with struc­tured data in the con­text of LLMs, and the flex­i­ble struc­ture is ide­al for answer­ing var­i­ous ques­tions, espe­cial­ly open-end­ed questions.” 

The func­tion­al­i­ty is in pre­view with select cus­tomers. The announce­ment builds on last year’s announce­ment of an Ana­lyt­i­cal Mod­el in SAP Data­s­phere for mul­ti­di­men­sion­al and seman­ti­cal­ly rich ana­lyt­i­cal mod­el­ing, allow­ing cus­tomers to bet­ter answer busi­ness questions.

SAP HANA Cloud Vec­tor Engine

With SAP HANA Cloud, its mul­ti-mod­el data­base man­age­ment sys­tem, SAP already pro­vides devel­op­ers with the abil­i­ty to build and deploy intel­li­gent data appli­ca­tions. A sin­gle, inte­grat­ed data plat­form that can store, access, and process dif­fer­ent types of data, SAP HANA Cloud serves as the data­base man­age­ment foun­da­tion of SAP Busi­ness Tech­nol­o­gy Platform. 

Vec­tor capa­bil­i­ties in SAP HANA Cloud will make it eas­i­er for AI mod­els to man­age unstruc­tured data and enhance inter­ac­tions between LLMs and orga­ni­za­tions’ mis­sion-crit­i­cal data. Cus­tomers will be able to sup­ply busi­ness con­text to ensure they are present in gen­er­a­tive AI outputs. 

Retriev­ing data and con­text in the native lan­guage of LLMs, vec­tors (long arrays of num­bers that give val­ue to a large num­ber of attrib­ut­es or dimen­sions) will allow SAP HANA Cloud users to gain more con­text on their busi­ness data. With this func­tion­al­i­ty, the SAP HANA Cloud data­base can be lever­aged to con­tin­u­ous­ly train LLMs under­neath orga­ni­za­tions’ gen­er­a­tive AI appli­ca­tions with up-to-date, rel­e­vant busi­ness infor­ma­tion. Accord­ing to SAP, this will sim­pli­fy inter­ac­tions with LLMs and encour­age devel­op­ers to build gen­er­a­tive AI into their own applications.

With new vec­tor data­base capa­bil­i­ties enhanc­ing the val­ue that cus­tomers can gain from its exist­ing mul­ti-mod­el capa­bil­i­ties, the SAP HANA Cloud vec­tor engine will addi­tion­al­ly allow cus­tomers to store and com­pare vec­tors in SQL, sim­pli­fy­ing data archi­tec­ture and secu­ri­ty. Cus­tomers will be able to com­bine spa­tial, graph, JSON, and rela­tion­al data with vec­tor queries. Addi­tion­al­ly, vec­tor data­bas­es can be used for retrieval-aug­ment­ed gen­er­a­tion (RAG), rec­om­men­da­tions, clas­si­fi­ca­tions, and clus­ter­ing, accord­ing to SAP.

This announce­ment builds upon SAP’s pre­vi­ous­ly announced vec­tor data store for SAP HANA Cloud. SAP HANA Cloud’s vec­tor engine is in pub­lic pre­view with select cus­tomers, with gen­er­al avail­abil­i­ty planned for late Q1 2024.

Inte­grat­ing SAP Data­s­phere with SAP Ana­lyt­ics Cloud 

To estab­lish SAP Data­s­phere more ful­ly as cus­tomers’ plat­form of choice for inte­grat­ing plan­ning process­es and tools, SAP will more close­ly inte­grate SAP Data­s­phere with SAP Ana­lyt­ics Cloud, includ­ing by mak­ing it pos­si­ble to deploy plan­ning mod­els from SAP Ana­lyt­ics Cloud into SAP Datasphere. 

Extend­ing SAP Ana­lyt­ics Cloud’s plan­ning func­tion­al­i­ty via SAP Data­s­phere will reduce the com­plex­i­ty and risk that cus­tomers face in stor­ing mul­ti­ple copies of data in dif­fer­ent sys­tems, bring­ing plan and actu­als data togeth­er to enable real-time steer­ing and fur­ther estab­lish a busi­ness data fab­ric across the SAP solu­tion suite. 

Inte­grat­ing SAP Ana­lyt­ics Cloud with Gen­er­a­tive AI

To accel­er­ate a holis­tic data and ana­lyt­ics strat­e­gy for cus­tomers, SAP is inte­grat­ing SAP Ana­lyt­ics Cloud, its solu­tion for busi­ness intel­li­gence and plan­ning, with its gen­er­a­tive-AI copi­lot, Joule, for data dis­cov­ery, dash­board cre­ation, and plan­ning mod­el main­te­nance. Joule will lever­age SAP HANA Cloud’s vec­tor engine to do so, ensur­ing that it main­tains vis­i­bil­i­ty of busi­ness con­text and logic.

Mean­while, through the ​“com­pass” fea­ture for SAP Ana­lyt­ics Cloud, build­ing upon pre­vi­ous Monte Car­lo sim­u­la­tions in exist­ing SAP tools, busi­ness users can more effi­cient­ly pre­dict or fore­cast busi­ness sce­nar­ios using enter­prise data, with­out advanced sta­tis­ti­cal skills. Busi­ness sim­u­la­tions can be per­formed direct­ly in SAP Ana­lyt­ics Cloud, with sce­nario mod­el­ing capa­bil­i­ties allow­ing for com­par­isons between assump­tions and improv­ing users’ abil­i­ty to ana­lyze rela­tion­ships between data.

A pre­view of SAP Joule func­tion­al­i­ty in SAP Ana­lyt­ics Cloud is planned for the end of Q2 2024, while a pre­view of SAP Ana­lyt­ics Cloud com­pass is planned for the end of Q3 2024.

Expand­ing the Open Data Ecosystem

SAP announced it will devel­op an AI gov­er­nance solu­tion in col­lab­o­ra­tion with Col­li­bra, which launched its own AI gov­er­nance solu­tion last month. By extend­ing its part­ner­ship with Col­li­bra (whose SAP and non-SAP meta­da­ta man­age­ment and gov­er­nance solu­tions were first inte­grat­ed with SAP Data­s­phere at launch) to include AI gov­er­nance, SAP will enable cus­tomers to see what data their var­i­ous AI mod­els have been trained on, to ensure that data sources remain trust­ed and accurate.

“We know that SAP data must be open and work with your exist­ing data infra­struc­tures,” said Mueller dur­ing SAP Data Unleashed. ​“That is why we announced the open data ecosys­tem last year, and we are con­tin­u­ing to deliv­er new capa­bil­i­ties with our partners.” 

Addi­tion­al­ly, SAP announced an inte­gra­tion part­ner­ship with Con­flu­ent. Through this part­ner­ship with Con­flu­ent for real-time ana­lyt­ics and deci­sion-mak­ing tools, cus­tomers can stream and con­nect busi­ness-ready SAP data to exter­nal data and appli­ca­tions in real time. ​“Cus­tomers can now send the right sig­nals to Kaf­ka-end points with SAP Data­s­phere data inte­gra­tion capa­bil­i­ties and soon repli­cate data from Con­flu­ent into SAP Data­s­phere as well,” Mueller said. 

For more infor­ma­tion on all the announce­ments, vis­it SAP.