Those who con­sid­er them­selves mem­bers of the pro­gram­ming cognoscen­ti know that soft­ware appli­ca­tion devel­op­ment is a slow and method­i­cal process of care­ful strate­gic code com­po­si­tion, builds, tests, debug­ging pro­ce­dures, and user accep­tance refinements.

If you recall SAP’s 2010 acqui­si­tion of Sybase, you may remem­ber Power­Builder being billed as a rapid appli­ca­tion devel­op­ment (RAD) tool. But even RAD isn’t always that fast in the grand scheme of things.

The Age of Arti­fi­cial Intel­li­gence (AI) Automation

How­ev­er advanced, auto­mat­ed, and accel­er­at­ed pre­vi­ous meth­ods of soft­ware devel­op­ment have been, the arrival of con­tem­po­rary arti­fi­cial intel­li­gence (AI) can now give pro­gram­mers the poten­tial to do things faster, smarter, more effec­tive­ly, and more intuitively.

AI and machine learn­ing (ML) tech­niques allow pro­gram­mers to dis­cov­er pat­terns, rep­e­ti­tions, and trends in data sets in an auto­mat­ed way. They can then apply this knowl­edge to map, mod­el, and ulti­mate­ly deploy func­tions in soft­ware appli­ca­tions that will be ful­ly cog­nizant of the under­ly­ing data trends gen­er­at­ed through user behav­ior. And increas­ing­ly machine behav­ior, too.

Find­ing Anom­alies and Pre­dict­ing Outcomes

AI and ML tech­niques also help pro­gram­mers dis­cov­er asso­ci­a­tions, con­nec­tions, and inte­gra­tion points between data sets in an auto­mat­ed way. This knowl­edge allows devel­op­ers to map, mod­el, and ulti­mate­ly deploy func­tions in the soft­ware appli­ca­tions they are build­ing that will be capa­ble of high­light­ing anom­alies, uncov­er­ing effi­cien­cies, and pre­dict­ing future outcomes.

How far any sin­gle programmer/​developer can go in terms of build­ing smarter AI-dri­ven apps will depend on the depth and qual­i­ty of the data pool they are able to use for their work.

Know­ing What We Don’t Know

Where AI and ML real­ly start to make a dif­fer­ence is when they show us not just the things we didn’t know, but the things we didn’t know we need­ed to know. As humans, we still must code our soft­ware appli­ca­tions based on what we think we need them to do for us. 

But as Meta­maven CTO Mariya Yao explains on Forbes, with AI and ML on board, a soft­ware engi­neer does not give the com­put­er rules for how to make deci­sions and take actions. ​“Instead,” she wrote, “[the engi­neer] curates and pre­pares domain-spe­cif­ic data that is fed into learn­ing algo­rithms, which are iter­a­tive­ly trained and con­tin­u­ous­ly improved. A machine learn­ing mod­el can deduce from data what fea­tures and pat­terns are impor­tant, with­out a human explic­it­ly encod­ing this knowledge.” 

What Yao is say­ing is that AI and ML out­puts can start to com­plete­ly sur­prise us and high­light appli­ca­tion fea­tures that we didn’t even know we need­ed to cre­ate, devel­op, main­tain, or extend. SAP echoes that in its def­i­n­i­tion of machine learn­ing, which states, ​“Machine learn­ing tech­nol­o­gy teach­es com­put­ers how to per­form tasks by learn­ing from data instead of being explic­it­ly programmed.”

School­ing Devel­op­ers in Arti­fi­cial Intelligence

How should ASUG Mem­bers be charg­ing their soft­ware appli­ca­tion devel­op­ment teams with AI- and ML-dri­ven advance­ments? SAP’s Ewan Maalerud has writ­ten on this sub­ject to explain a few use cas­es for these advance­ments, includ­ing how AI enhances what you can do with ana­lyt­ics in SAP Ana­lyt­ics Cloud.

Maalerud points to SAP Ana­lyt­ics Cloud’s abil­i­ty to per­form as an ana­lyt­ics plat­form that uses AI in its pre­dic­tive mod­el­ing. It uses rich data visu­al­iza­tions with an intu­itive user inter­face. Devel­op­ers can chan­nel this AI brain­pow­er into appli­ca­tions so they can con­vey a rich­er con­tex­tu­al under­stand­ing and sit­u­a­tion­al aware­ness of any firm’s activ­i­ties to help employ­ees make informed decisions.

A Jump-Start for AI and ML Projects

This dis­cus­sion would be incom­plete if we did not men­tion the SAP Leonar­do Machine Learn­ing Foun­da­tion. Although the name may sound like a char­i­ta­ble foun­da­tion, it’s actu­al­ly an offer­ing to help advance machine learn­ing at orga­ni­za­tions through ready-to-use ser­vices and mod­els avail­able through APIs and web ser­vices. If you’re think­ing of bring­ing AI or ML into your process­es, this can act as a jump-start. 

Devel­op­ers can sim­ply point their appli­ca­tions in the right direc­tion to APIs con­nect­ed up with ser­vices that can (for exam­ple) detect and iden­ti­fy objects in pic­tures, find sim­i­lar images and text con­tent, or extract key­words from nat­ur­al lan­guage text.

Machine Learn­ing Mod­els to Tune Up

Accord­ing to SAP, ​“Besides using pre­trained ML ser­vices you can also deploy cus­tom ML mod­els or tune exist­ing mod­els with your own train­ing data. This allows you to eas­i­ly serve cus­tomized ML mod­els for crit­i­cal busi­ness process­es in a scal­able and secure manner.”

AI for pro­gram­mers inside or out­side of an SAP envi­ron­ment brings automa­tion pos­si­bil­i­ties, con­tex­tu­al knowl­edge advan­tages, appli­ca­tion effi­cien­cy, and deci­sion sup­port. Sure­ly that’s knowl­edge worth gath­er­ing and apply­ing today.

If you’re plan­ning to attend SAP­PHIRE NOW and ASUG Annu­al Con­fer­ence, don’t miss the day of learn­ing about SAP Leonar­do. There will be an ASUG Pre-Con­fer­ence Sem­i­nar on Busi­ness Process Inno­va­tions for the Intel­li­gent Enter­prise and all things relat­ed to SAP Leonar­do. Reg­is­ter today and save your spot.