Alleviating the organisational tension blocking AI success

The finan­cial ser­vices indus­try has evolved rapid­ly since the glob­al finan­cial cri­sis of 2008 spawned the rise of fin­techs and, with them, a cus­tomer base with increas­ing­ly demand­ing dig­i­tal expec­ta­tions. The pan­dem­ic has accel­er­at­ed dig­i­tal behav­iours even fur­ther, see­ing tech­nol­o­gy leap over brand trust as the key bat­tle­ground on which new cus­tomers are won.

Specif­i­cal­ly, arti­fi­cial intel­li­gence (AI) is becom­ing the key dif­fer­en­tia­tor. With dig­i­tal hav­ing thor­ough­ly eclipsed phys­i­cal branch­es as the dom­i­nant medi­um of engage­ment for finan­cial ser­vices, com­pa­nies must now focus on per­son­al­is­ing their cus­tomer expe­ri­ence through the likes of chat­bots, vir­tu­al assis­tants and video bank­ing. If they are not able to pro­vide these kinds of AI-enabled expe­ri­ences, their cus­tomer ser­vice will become infe­ri­or.

“If you want to dri­ve cus­tomer life­time val­ue and reten­tion, cross-sell prod­ucts and deliv­er bet­ter cus­tomer ser­vice, it has to be in a dig­i­tal envi­ron­ment and ecosys­tem, and to be best in class, it must be AI-infused,” says Kevin Levitt, indus­try busi­ness devel­op­ment lead for finan­cial ser­vices at NVIDIA. “Finan­cial ser­vices firms undoubt­ed­ly now see the impor­tance of AI to their com­pet­i­tive posi­tion, and the need, there­fore, to set an enter­prise AI strat­e­gy.”

Thanks to the capa­bil­i­ties of com­pa­nies like NVIDIA and Hewlett Packard Enter­prise (HPE), the com­put­ing pow­er required to deploy AI in enter­prise envi­ron­ments has caught up with the the­o­ry, accel­er­at­ing AI projects. While some are still only dis­cussing AI, and at the oth­er end of the scale it is almost per­va­sive in some star­tups, most com­pa­nies have reached a stage of hav­ing now enjoyed suc­cess with sev­er­al small­er AI deploy­ments and wish to scale those quick wins into some­thing larg­er.

There are, nat­u­ral­ly, tech­nol­o­gy chal­lenges involved with scal­ing up AI in an enter­prise, not least estab­lish­ing a coher­ent data strat­e­gy to under­pin deploy­ments, but the great­est chal­lenges are organ­i­sa­tion­al. A tri­fec­ta of per­sonas is key to the suc­cess of any AI deploy­ment – IT, data sci­en­tists and line of busi­ness – but con­trast­ing roles and objec­tives can often mean they are not only speak­ing dif­fer­ent lan­guages but that there is active ten­sion.

The effi­cien­cy of the data sci­en­tists, a high­ly cost­ly resource, is crit­i­cal to the suc­cess of deploy­ments. If the data sci­ence group can­not work effec­tive­ly between IT and the line-of-busi­ness project own­ers, the project is like­ly to floun­der. They require a strong oper­a­tional path that enables them to use the tools they want, how they want, and access the resources they need, with­out usu­al IT block­ers like rais­ing a tick­et or wait­ing for access.

“Col­lab­o­ra­tion is vital,” says Adri­an Lovell, chief tech­nol­o­gy offi­cer, finan­cial ser­vices indus­try, at HPE. “If you’re not on top of that sort of dynam­ic, in the best case it’s going to take you sig­nif­i­cant­ly longer to derive any val­ue from your efforts, and in the worst case you’re not going to get any­thing out of the lab and it’s just going to die there as a sci­ence project. You’ve absolute­ly got to find a way to break through the fric­tion and facil­i­tate the col­lab­o­ra­tion that is required.

“It’s about under­stand­ing that nobody wants to cause a prob­lem. The IT peo­ple may sim­ply feel pro­tec­tive over what they do for a liv­ing and don’t like the idea that some­body else may want to try and change that. The data sci­en­tists may have come from acad­e­mia or anoth­er indus­try where they were able to do every­thing they always want­ed to do and are now adjust­ing to a struc­tured, cor­po­rate envi­ron­ment with­out as many of those free­doms. And then there are the prod­uct own­ers say­ing, ‘I’m try­ing to make mon­ey here, just get it done.’”

Amidst these organ­i­sa­tion­al ten­sions there is also the creep­ing issue of shad­ow AI, where­by the pres­sure on lines of busi­ness to lever­age AI-enabled appli­ca­tions has result­ed in siloed, frag­ment­ed imple­men­ta­tions, some in the cloud and oth­ers on-premise. There is no AI infra­struc­ture at scale across the enter­prise, leav­ing finan­cial ser­vices com­pa­nies strug­gling to get to mar­ket with the AI-enabled ser­vices that will help dif­fer­en­ti­ate them.

“All these chal­lenges are bub­bling to the sur­face,” Levitt adds. “That’s the con­flu­ence of what hap­pens when these three impor­tant pieces of the puz­zle are not organ­ised and talk­ing to each oth­er. Ulti­mate­ly it sur­faces a bro­ken AI enter­prise because there is no strat­e­gy to go with it. That’s where I would say there’s a lot of ten­sion but also oppor­tu­ni­ty in terms of the con­ver­sa­tions that we’re hav­ing across the exec­u­tive suite today with­in finan­cial ser­vices.”

Finan­cial ser­vices organ­i­sa­tions are look­ing to part­ner with lead­ers in this space to help them under­stand how to deploy effec­tive AI infra­struc­ture at scale to get the most val­ue out of one of their most val­ued resources: their data sci­en­tists. Many are turn­ing to a part­ner­ship between HPE and NVIDIA, which togeth­er can pro­vide an enter­prise AI infra­struc­ture at scale that’s full stack, from the hard­ware up to the appli­ca­tion suite, enabling com­pa­nies to oper­a­tionalise AI-enabled appli­ca­tions for any vari­ety of use cas­es.

Wait­ing for a mod­el to train for days or poten­tial­ly a week is an incred­i­ble waste of a valu­able resource in a data sci­ence team. The com­bined HPE and NVIDIA solu­tions help finan­cial ser­vices firms wher­ev­er they are in their AI matu­ri­ty, from those already see­ing the ben­e­fits and want­i­ng to scale that capa­bil­i­ty, to the lag­gards which are now real­is­ing they can’t com­pete in the mar­ket­place unless they devel­op a for­mi­da­ble AI enter­prise strat­e­gy. Procur­ing solu­tions as-a-ser­vice can also assist the chal­lenges between the tri­fec­ta, with HPE pro­vid­ing all solu­tions as a ser­vice through HPE Green­lake.

“It’s all about the part­ner­ship when you are tack­ling these sorts of chal­leng­ing prob­lems,” says Lovell. “I don’t think any­body would attempt to try and do this by them­selves. We like to work with our friends at NVIDIA because we believe in bring­ing what the best of breed is to our finan­cial ser­vices cus­tomers to ulti­mate­ly try and address those fun­da­men­tal prob­lems.

“We’ve been work­ing with NVIDIA to help the tri­fec­ta of per­sonas involved in AI deploy­ments under­stand each oth­er’s point of view, because that’s real­ly not a tech­nol­o­gy prob­lem. Tech­nol­o­gy can help the prob­lem but fun­da­men­tal­ly it’s about under­stand­ing each oth­er’s point of view and see­ing, frankly, that every­body needs to work togeth­er or noth­ing will be deliv­ered. That’s all that mat­ters to the busi­ness at the end of the day: get­ting it deliv­ered.”

For more infor­ma­tion, vis­it hpe.com/uk/getahead

Pro­mot­ed by HPE and NVIDIA