Trends and the future of AI: From opti­miz­ing costs to plan­ning your Christ­mas party

At first glance, it may appear that AI is only con­nect­ed with pro­gres­sive­ly mind­ed star­tups. How­ev­er, the real­i­ty is that even large cor­po­ra­tions use it to a great extent. Chat­bots have pri­mar­i­ly tak­en root with them, like at JPMor­gan Chase, where chat­bots assist finan­cial advi­sors in com­mu­ni­cat­ing effec­tive­ly. Con­sul­tants at Mor­gan Stan­ley then use them for edu­ca­tion­al pur­pos­es. In many com­pa­nies today, they fill the roles of for­mer work posi­tions – at Brex, where they devel­op soft­ware for mon­i­tor­ing costs, they intro­duced a finan­cial­ly lit­er­ate AI assis­tant specif­i­cal­ly designed for finan­cial direc­tors and accountants. 

Last year, how­ev­er, star­tups and prod­uct-ori­ent­ed com­pa­nies took it a step fur­ther, not lim­it­ing them­selves to chat­bots only. A good exam­ple is the com­pa­ny Sales­force, which is chang­ing the world of CRM with the help of an intel­li­gent sys­tem for plan­ning meet­ings and per­son­al­is­ing fol­low-ups. Also, last year, Adobe pushed the bound­aries of cre­ativ­i­ty by fill­ing and mod­i­fy­ing images in Pho­to­shop, while Notion com­plete­ly changed how we work with a cor­po­rate knowl­edge base by sum­maris­ing and gen­er­at­ing texts. If you work in Notion, then you’ve sure­ly noticed these func­tions. One such fea­ture (a rather con­tro­ver­sial one, if you ask me) is the ​“make longer” option, which arti­fi­cial­ly extends texts. On the one hand, we’re using AI in droves to sum­marise long texts so we don’t have to read them and sim­pli­fy work­ing with them, and on the oth­er, tools are being made to arti­fi­cial­ly inflate them even more. In the future, I hope we choose to steer away from devel­op­ing such prod­ucts. Oth­er­wise, we’ll go down the road of cre­at­ing a world where one AI writes a text just so anoth­er AI can read it.

Investors are also expect­ing big things from gen­er­a­tive AI. Its progress last year act­ed as an impulse for VISA, which launched a new cap­i­tal fund. This fund amount­ed to 100 mil­lion dol­lars for star­tups that use gen­er­a­tive AI. VISA will most like­ly use this mon­ey to strength­en autonomous agents as much as pos­si­ble, con­nect sys­tems, and help com­pa­nies devel­op more valu­able prod­ucts. Although every­thing main­ly revolves around ​“new” gen­er­a­tive AI, it’s impor­tant to observe that invest­ments in the sec­tor are also push­ing for­ward the clas­sic AI meth­ods, like pre­dic­tions, data analy­sis, sched­ul­ing, and machine learn­ing as a whole. 

The pre­dic­tions made by these giants are quite pos­i­tive. One such per­spec­tive is offered by the con­sul­ta­tion com­pa­ny McK­in­sey, for exam­ple. It esti­mates that AI could increase pro­duc­tiv­i­ty next year in the finan­cial sec­tor by near­ly 5%. That would equate to annu­al com­bined sav­ings of hun­dreds of bil­lions of dol­lars. I myself believe that gen­er­a­tive AI’s full poten­tial has yet to be achieved. If I were to liken this path to a hike up a moun­tain, then we’ve only tak­en a few steps from the base camp – today, we’re look­ing at Chat­G­PT the same way peo­ple looked at the first per­son­al com­put­ers in 1991 when only a few peo­ple were active­ly using them. A long evo­lu­tion awaits these tools, and we still have so much to discover. 

This year, I expect AI to become some­thing even more com­pe­tent. It will solve its tasks more pre­cise­ly, effec­tive­ly, and, most impor­tant­ly, inde­pen­dent­ly. It’s quite like­ly that autonomous AI agents will turn our way of work­ing on its head. I assume that at Applift­ing it will iden­ti­fy places in our bud­get where we can save mon­ey, over­haul our capac­i­ty, and even plan the annu­al Christ­mas par­ty. While the tool takes over our hours-long dai­ly rou­tines, we can instead focus on the work we enjoy and direct­ly gen­er­ate income from. Plus, the best sys­tems will cre­ate their own agents to fill par­tial roles. The tool Auto-GPT is a good exam­ple of this, giv­en it has made incred­i­ble progress just this past year. 

The con­stant cheap­en­ing of the APIs that help us use these AI tools is fun­da­men­tal to this growth. We can also expect a rapid com­modi­ti­sa­tion of the mar­ket – it’s safe to say that soon it won’t mat­ter if a task is per­formed by a prod­uct of Ope­nAI, Google, or anoth­er com­pa­ny. The price will decide. 

When it comes to gen­er­a­tive AI, last year was in the spir­it of appli­ca­tions. We’re still wait­ing for a sim­i­lar boom in the area of hard­ware. More­over, the mobile phone mar­ket is a mas­sive yet undi­vid­ed piece of the pie that is up for grabs for any com­pa­ny. And who knows? Maybe they’ll cre­ate a com­plete­ly new one in the near future. 

The first swal­low fly­ing in this direc­tion was sure­ly AI Pin, cre­at­ed in the work­shop of the com­pa­ny hu.ma.né. At first glance, it appeared quite promis­ing – a small box you can pin to your jack­et and receive answers to any ques­tions you have. It sound­ed like a dream, which, unfor­tu­nate­ly, couldn’t become a real­i­ty. The respons­es from AI Pin were too sim­ple, some were even com­plete­ly off-top­ic. The main prob­lem, how­ev­er, was that nobody, includ­ing the founders, man­aged to effec­tive­ly explain what AI Pin is actu­al­ly good at. And I won’t even speak to the absurd 700-dol­lar price tag. 

After hu.ma.né’s fail­ure, oth­er inter­est­ed projects start­ed to appear. For instance, the com­pa­ny Rab­bit intro­duced the R1 device, which saw great mar­ket suc­cess from its very first week. It made a remark­able 10 mil­lion dol­lars just in pre-sales. Hav­ing paid atten­tion, these founders pre­pared a pow­er­ful AI mod­el as well as an overview of poten­tial use cas­es. That’s why cus­tomers can use R1 for simul­ta­ne­ous trans­la­tions and order­ing food and taxis. There are also anoth­er 15 func­tions that Rabbit’s plan­ning to add. Unlike AI Pin, R1 is more akin to the first push-but­ton iPod, mak­ing it eas­i­er for users to use. And the more rea­son­able 200-dol­lar price is anoth­er bonus. 

Maybe you’re think­ing to your­self, ​“Where’s Apple?”. Despite hav­ing missed the boat a bit with its voice assis­tant Siri, it cer­tain­ly isn’t behind in AI hard­ware devel­op­ment. In fact, we can expect big things from Apple (like usu­al) that would be out­side oth­er brands’ bud­gets – for exam­ple, I can pic­ture Siri being able to con­trol oth­er appli­ca­tions on her own. We can fur­ther our assump­tions by look­ing at Apple’s acqui­si­tions in past years. While it’s pur­chased twen­ty star­tups focus­ing on AI devel­op­ment since 2017, Microsoft only has twelve at this point, Meta owns eleven, and Alpha­bet (Google) only eight. Whether we’re talk­ing about soft­ware or hard­ware, I think it’ll be Apple with its fin­ger on the pulse. 

Is it even pos­si­ble to secure a seat in such a fast-fill­ing are­na? I cer­tain­ly believe so. You need to be able to see around the cor­ner, how­ev­er. Inno­va­tion doesn’t mean copy­ing what oth­ers are doing. You need to devel­op what­ev­er exceeds cur­rent pos­si­bil­i­ties and do it fast because, before you know it, your com­peti­tor will have the same idea. 

But if you’re look­ing for tips on how to sim­pli­fy your work today using AI, then learn to work with chat­bots. There are so many to choose from, and thanks to Ope­nAI and the GPT Store, the list is ever-grow­ing. Remem­ber, there’s no time to waste. And if you already use gen­er­a­tive AI in your com­pa­ny, con­tin­ue to do so. But don’t rest on your lau­rels – keep up with the news, and don’t be afraid to exper­i­ment. Feel free to write to us if you’re unsure where to start with these tools. At Applift­ing, we work with AI dai­ly, and we’d hap­pi­ly share our expe­ri­ence with you. 

This arti­cle was orig­i­nal­ly writ­ten in Czech for ​“IT Sys­tems” mag­a­zine.