Kungani i-Salesforce ne-Microsoft Balwela Ikusasa Lama-AI Agents

Uyazi ukuthi ubuchwepheshe obusafufusa buba ngokoqobo uma amafemu e-Big Tech ezama ukuba ngabanikazi babo. Yilokho okwenzekayo njengoba i-Salesforce izama ukubamba ubuholi bemakethe yama-agent e-AI ezimele.

Abenzeli be-AI amasistimu ahlakaniphile aklanyelwe ukwenza imisebenzi ngokuzenzakalelayo. Emcimbini wayo waminyaka yonke we-Dreamforce, i-Salesforce yethula i-AgentForce, i-ejenti yayo ye-AI ezimele yendawo yokusebenza . Ngemuva kwalokho, i-CEO ye-Salesforce uMarc Benioff waba matasa kakhulu ehlasela inani lamamodeli ezilimi ezinkulu (LLMs) futhi ethwebula iMicrosoft ngesilinganiso esihle. Ngokwazi:

• Muva nje uthe abasizi be-AI abakhiqiza i-LLM, njengeMicrosoft Copilot, bagcwele ngokweqile. Ku-Rapid Response podcast, uthe, “Kungenzeka ukuthi sezwa kulaba baphristi nabapristikazi be-AI balezi zinkampani eziyimodeli ye-LLM kanye ne-Microsoft nabanye mayelana ne-AI yokwelapha umdlavuza nokushintsha kwesimo sezulu, futhi sonke kufanele sixhume kulezi zitshalo zamandla enuzi ukuze sithole. lezi zikhungo zedatha. Akukho kulokhu okuyiqiniso.”

• Futhi Ngemuva kokuthi iMicrosoft ithe izovumela izinhlangano ukuthi bakhe awabo ama-autonomous AI agents ngaphakathi kwesiteji se-Copilot Studio, u-Benioff ukhiphe a ukwehliswa okunonya ku-X : "I-Microsoft iqamba kabusha u-Copilot ngokuthi 'ama-ejenti'? Leyo imodi yokwethuka. Masibe ngokoqobo — I-Copilot iyi-flop ngoba i-Microsoft ayinazo idatha, imethadatha, namamodeli okuphepha ebhizinisi ukudala ubuhlakani bangempela benkampani."

Kungani i-CEO ematasa yenkampani yomhlaba wonke ye-Big Tech ebiza izigidigidi ingathuka kuma-LLM nakuMicrosoft?

Ngoba uBenioff uyazi ukuthi abenzeli be-AI babaluleke kangakanani. Lokhu kusho:

• Ufuna ukususa ingxoxo emayelana ne-AI kuma-LLM futhi iye kubenzeli be-AI.

• Ubona ukungena kweMicrosoft kubenzeli be-AI njengosongo. Ukube akakwenzi lokho, ubengasinaka isimemezelo seMicrosoft.

  
  

Ngokwenza kanjalo, wagcizelela ukubaluleka kwama-agent e-AI. Futhi uthuthukise iMicrosoft .

Ukubaluleka kwama-AI Agents

Imakethe yama-autonomous AI agents ikhula ngokushesha. Imakethe yama-agent e-AI kulindeleke ukuthi inwebe isuka ku-$5.1 billion ngo-2024 iye ku-$47.1 billion ngo-2030, okubonisa izinga lokukhula elihlanganisiwe lonyaka lama-44.8% . I-Salesforce ifuna ucezu lwaleyo makethe.

Kodwa izinombolo aziwuthathi kahle amandla azo. Eqinisweni, ama-agent e-AI abe yisihloko esishisayo esimela ukweqa okubalulekile ku-AI ecosystem. Ngokungafani nezinhlelo ze-AI zendabuko ezisabela emiyalweni, ama-agent e-AI angaphishekela izinjongo ngokuqhubekayo, afunde kokuhlangenwe nakho kwawo, futhi aze ahlanganyele namanye ama-ejenti nabantu ngezindlela eziyinkimbinkimbi.

Ngenxa yalokho, abenzeli be-AI bangase benze imisebenzi evamile, eyisicefe abantu abangafuni ukuyenza - njengokufaka idatha, ukuphatha amakhalenda, ukwenza ngokuzenzakalelayo ukudalwa kwemibiko, nokunye okuningi. Ngamanye amazwi, lokho i-AI ebilokhu ikuthembisa ngaso sonke isikhathi.

Lokhu kuzimela kuye kwavusa kokubili injabulo novalo.

I-Venture capitalist uJeremiah Owyang ucabange ukuthi ama-AI agents azokwenza thuthukisa amawebhusayithi ngoba amabhizinisi azodinga ukuthola ukuthi angasetshenziswa kanjani kabusha izingosi zawo kubenzeli be-AI abenza ukusesha kuwebhu esikhundleni sabantu. UChris Mattmann, owayeyisikhulu esiphezulu se-NASA JPL kwezobuchwepheshe kanye ne-innovation, uveza ukukhathazeka kwabaningi ngokubuza ukuthi ilungele umhlaba lapho i-AI isebenza khona ngaphandle kokuqondisa komuntu .

U-Gartner uhlonze i-“Agentic AI” njengethrendi ephezulu yezobuchwepheshe ka-2025 , egcizelela amandla ayo okuguqula ukusebenza kwebhizinisi. U-Gartner ubikezela ukuthi ngo-2028, ama-33% ezinhlelo zokusebenza zesofthiwe yebhizinisi azofaka i-AI ye-agent, evumela u-15% wezinqumo zomsebenzi wansuku zonke ukuthi zenziwe ngokuzimele. Nokho, uGartner uxwayisa ngesidingo sokubusa ukuze kuvinjwe izingozi ezifana nokuchema kanye nezindaba zobumfihlo. UGartner ugcizelela ukubaluleka kokusungula imigwaqo yokuqapha ukuze kuqinisekiswe ukuthi i-AI ihambisana namagugu nezinjongo zomuntu.

UMarc Benioff ngokumelene neMicrosoft

Lokhu kusibuyisela ku-Marc Benioff. Usewafundile amahlamvu etiye. Ubona ama-agent azimele njengekusasa, futhi ubeke ukubheja kwakhe ngokufanele. I-Salesforce izishumekile kuyo 150,000 izinhlangano . Uma ushumekwe ngokujulile kuyo yonke inkampani yaseMelika, umkhiqizo wakho uba i-Trojan Horse yobuchwepheshe obusha. Isibonelo, i-Salesforce AgentForce izosebenzisa ama-agent e-AI azimele athuthukisa isevisi yamakhasimende, ukuthengisa, ukumaketha, kanye nemisebenzi yezentengiselwano.

Udinga ukwenza ngokuzenzakalelayo imisebenzi yenjwayelo efana nokuphendula imibuzo yamakhasimende, imikhondo yokuthengisa efanelekayo, nokwenza imikhankaso yokumaketha ibe ngcono? Sawubona, kukhona i-ejenti yalokho - i-AgentForce. Udinga indlela esebenza kahle kakhulu yokuthola imininingwane eqhutshwa idatha kwa-Salesforce? Nokho, i-AgentForce isebenzisa i-Atlas Reasoning Engine ukuze ihlaziye idatha, yenze izinqumo, futhi yenze imisebenzi ngokuzenzakalelayo.

IMicrosoft ayikuthathi konke lokhu ukuhlala phansi. Ngo-Okthoba 21, iMicrosoft kumenyezelwe :

• Ikhono lokudala ama-agent azimele nge-Copilot Studio.
• Ukwethulwa kwama-ejenti amasha ayi-10 azimele ku-Dynamics 365 ukuze kwakhiwe amandla okuthengisa, isevisi, ezezimali kanye nemisebenzi ye-supply chain.

Nansi indlela eqondile iMicrosoft echaze ngayo kubhulogi yayo umthelela ongaba khona wama-agent we-AI:

I-Copilot ingumsizi wakho we-AI - ikusebenzela - futhi I-Copilot Studio ikuvumela ukuthi udale kalula, uphathe futhi uxhume ama-ejenti ku-Copilot. Cabanga ngama -agent njengezinhlelo zokusebenza ezintsha zomhlaba one-AI . Yonke inhlangano izoba neqoqo lezinkanyezi zama-ejenti - kusukela ekushesheni nasekuphenduleni kuya ekuzilawuleni ngokugcwele. Bazosebenzela umuntu oyedwa, iqembu noma umsebenzi ukuze baqhube futhi bahlele inqubo yebhizinisi. I-Copilot yindlela ozosebenzisana ngayo nalawa ma-ejenti, futhi azokwenza yonke into kusukela ekusheshiseni ukukhiqiza okuholayo nokucubungula ama-oda okuthengisa kuye ekwenzeni ngokuzenzakalelayo uchungechunge lwakho lokuhlinzeka.

Empeleni, iMicrosoft izolwa neTrojan Horse yayo: amaphesenti angama-60 eFortune 500 asebenzisa iMicrosoft 365 Copilot (ngeMicrosoft ngayinye).

Isikhulu esiphezulu seMicrosoft sigweme impi yamagama noBenioff. Esikhundleni salokho, uthembele kuJared Spataro ukuthi aphendule amazwi ka-Benioff aphikisayo. utshela i-CRN , “Wonke amakhasimende asendaweni ehlukile ohambweni lwawo, kodwa sekukonke sizwa okuhluke kakhulu ku-Copilot wethu kumakhasimende e-Microsoft 365 . . . Ngekota edlule iyodwa, sibone ukwanda kwamakhasimende ngamaphesenti angaphezu kuka-60, futhi abasebenzisi bansuku zonke baphindeke ngaphezu kokuphindwe kabili – okuyinkomba ecacile yenani le-Copilot emakethe.”

UMarc Benioff ngokumelene nama-LLM

Angazi ukuthi izohamba kanjani impi phakathi kweSalesforce neMicrosoft. Nokho, ngikholwa ukuthi u-Benioff akalungile ukuhlasela izinhlelo zokusebenza ze-AI ezisekelwe ku-LLM ezisekelwe ekwazisweni kwabantu. I-ChatGPT isineminyaka emibili kuphela isemakethe, futhi kakade imikhiqizo ye-AI ekhiqizwayo engekho eshalofini esekelwe kuma-LLM ishintsha izimpilo zethu. Ngaphezu kwalokho, sisesezinsukwini zokuqala zamabhizinisi akha amamodeli awo akhiqizayo e-AI ngokusekelwe kudatha yangasese.

Cabangela izimboni ezimbili ezisithinta sonke: izinsizakalo zezezimali nokunakekelwa kwezempilo.

• Amasevisi ezezimali: Ama-LLM ashintsha indlela amasevisi ezezimali enza ngayo umsebenzi wawo, amafemu afana ne-BlackRock ewasebenzisa ukuhlaziya idatha yemakethe nokuhlonza amathuba okutshala izimali. Ithimba le-BlackRock elisebenzayo lokulingana elisebenzayo lisebenzisa ama-LLM ukuze licubungule amasethi edatha, lidalule amaphethini izindlela ezivamile ezingase zigeje. I-Hedge fund i-Bridgewater Associates, ephethe iphothifoliyo engu-$19 billion, isebenzisa ama-LLM ku-AIA Platform yayo ngemisebenzi eyinkimbinkimbi njengocwaningo lwemakethe.
  
• Ukunakekelwa kwezempilo: izibhedlela zithola indlela yokusebenzisa ama-LLM ukuze kuthuthukiswe ukusebenza kahle kulo lonke ibhodi, kusukela ekuthuthukiseni ukuthathwa kwegumbi lezimo eziphuthumayo kuya ekuthuthukiseni ukunakekelwa komtholampilo. Isibonelo, i-HCA Healthcare eTennessee ihlanganisa ama-LLM ezibhedlela zayo, igxile ekuthuthukiseni imisebenzi edla isikhathi njengokubhala amanothi asuka ezingxoxweni neziguli ngesikhathi sokuvakasha.

Izibonelo ezingenhla zibonisa indlela ubuchwepheshe obushintsha ngayo umhlaba: umsebenzi owodwa ngesikhathi. Ngezinye izikhathi ngezindlela ezibonakalayo. Imvamisa ngemuva kwezigcawu.

I-Benioff ilungile ekucindezeleni amafemu ezobuchwepheshe ukuthi athuthukise amamodeli e-AI asekelwe ku-LLM. Benza amaphutha futhi babhekane nokugxekwa ngezindaba ezinjengokuchema kanye ne-copyright yobuhlakani. Kodwa ngakolunye uhlangothi, kanjalo nama-AI agents. Kubalulekile ukubaqeqesha kahle futhi ugcine abantu benolwazi lokulungisa amaphutha futhi uvimbele ukuchema. Kodwa ushaya indiva iqiniso lokuthi ama-LLM afunda emaphutheni awo. Bayathuthuka. Ngokushesha. Uma kukhona, umthelela walokho ama-LLM angakwazi ukukwenza nazokwenza ubukelwa phansi.