Umsebenzi wakho omtsha kukukhwela iiarhente ze-AI: IiNkampani zoMthonyama ze-AI zisebenza njani ngokwenene

In This Article

Ababhalisi abathandekayo, ndichithe iinyanga ezimbalwa ezidlulileyo ndixoxa neenkokeli kwiinkampani ze-AI. Ngoku ndiqinisekile ukuba: Ukukhwela kunye nokulawula ii-arhente ze-AI NGUmsebenzi, kungakhathaliseki ukuba yintoni umsebenzi wakho.Kule ngqungquthela yasimahla ye-dive, ndifuna ukwabelana ngeendlela ezintathu zeenkampani ze-AI-i-Linear, i-Ramp, kunye ne-Factory - eziwubeka ngayo lo mgaqo. Nazi ezinye iingcaphuno ezivela kumntu ngamnye: uNan Yu (iNtloko yeMveliso, iLinear): “Uya kuba namalungu eqela le-AI onokuwabela imisebenzi kwaye uthethe ngendlela othetha ngayo nabantu.” UGeoff Charles (CPO, Ramp): “Ukuba awusebenzisi iKhowudi kaClaude, nokuba yintoni indima yakho, mhlawumbi awusebenzi kakuhle.” U-Eno Reyes Ucofelwe phambili Imveliso, Ulawulo lwe-CTO, idata uhlalutyo, nangakumbi kwizakhono ezinokuphinda zisetyenziswe nawuphi na umqeshwa anokuthi azibongoze. ”Funda kujongwe lwangaphakathi kwindlela inkampani nganye ye-AI esebenza ngayo.Linear: Ukwenza ii-arhente ze-AI zodidi lokuqala lweqelaUngakwazi @mention i-arhente ukuyila kunye nokwabela umba kwenye i-arhenteLinear's approach to agents is shaped by their product. UNan (iNtloko yeMveliso kaLinear) ukholelwa ukuba: Iiarhente kufuneka zibe ngabasebenzi abakumgangatho wokuqala. Kuya kufuneka ukwazi ukuzongeza kwiiprojekthi, uzinike imiba, kwaye ukhankanye kwizimvo. Nangona kunjalo, uNan ukholelwa ukuba umntu uhlala ephendula ngesiphumo. Nantsi indlela i-Linear eyakha ngayo iimveliso kunye neearhente kwinyathelo ngalinye: Qonda ingxaki. Iiarhente zifunde kwaye zishwankathele yonke incoko yabathengi ukusuka kwi-Intercom, iZendesk, kunye neGong. Benza imiba ngokuzenzekelayo, bayikhuphe ngokuchasene ne-backlog, kwaye bayinike iqela elifanelekileyo. Chonga isisombululo. Kuba iiarhente zinokufikelela kwiincoko zabathengi, zinokukunceda ukuba uphindaphinde into ethile ngokutsala ulwazi oluxhaswa yidatha kumajelo amaninzi. Yenza isicwangciso. Iiarhente zinokwaphula i-spec yakho ibe ngamatikiti ekhonkrithi kwaye ziwahambise kumaqela afanelekileyo ngokuzenzekelayo. KwaLinear, iiarhente ngoku zenza uninzi lwamatikiti. Umsebenzi womntu kukuphonononga umsebenzi wabo kunye nokulungelelanisa imeko ngokuhamba kwexesha.Execute. Iibugs kunye neempawu ezincinci zinikwa ngokuthe ngqo kwiiarhente ezinjengeCodex kunye neCursor. Kwiimpawu ezinzima, iinjineli zisungula iKhowudi yeClaude kwaye zisebenzise i-Linear MCP ukutsala umxholo opheleleyo wemiba.Ukusuka kuNan: Kuvakalelwa kukuba i-scope yento ii-arhente ezinokusingatha ngayo ukwandisa ikota nganye. Iimodeli ezintsha kunye neentambo zokubopha zityhala umda ukusuka kulungiso olulula ukuya kwiiprojekthi ezintsonkothileyo.Ufuna ukwakha ngeearhente ezifana neLinear? Nanga amanyathelo ama-4 asebenzayo athe uNan abelane ngawo malunga nento enokwenziwa liqela lakho namhlanje:Wonke umphuhlisi kufuneka angagqibekanga kwisixhobo sokukhowuda esikhokelayo. Eli lelona nyathelo lokuqala lilula. Nikezela ngesixhobo esisemthethweni (iCursor, Claude Code, okanye iCodex) kwaye uyilawule ukuze ukwazi ukubona ukusetyenziswa.Ukongeza nge-async cloud coding agent. Iiarhente zangasemva zeAsync zinokudubula olunye utshintsho oluncinci kunye nokulungiswa kwebug. I-cursor kunye ne-Devin baneminikelo emihle apha.Ukugxininisa ukuba abaqulunqi kunye nee-PM basebenze ngokuthe ngqo kwi-codebase. Ii-Agents ezifana noClaude zivula indlela ephantsi ye-friction ye-PM kunye nabaqulunqi ukuba benze utshintsho ngokuthe ngqo kwi-codebase. Wonke umntu kufuneka azame ukuba ngumakhi.PMs kunye nabathengisi kufuneka bangagqibekanga kwi-interface ye-AI. Le misebenzi kufuneka yenze i-80-100% yomsebenzi wabo ngokusebenzisa i-interface yengxoxo - nokuba yi-Claude, i-ChatGPT, i-Notion AI, okanye into efanayo. I-Nan ibona ikamva apho abantu baya kusebenzisana kunye nee-agent kwinqanaba elikhethekileyo - lichaza oko kufuneka kwakhiwe kwaye kutheni - kwaye emva koko udlulisela izinto kwii-arhente ukuphatha yonke into ephantsi.I-Ramp: Ukulinganisa ubuchule be-AI kumanqanaba e-4Iramp igxotha ngokukhawuleza abasebenzi babo ukuba babe yi-AI-native kumanqanaba e-4Ukuba i-Linear ibonisa indlela yokwenza ii-agent zibe yi-agent-i-agent ye-agent-i-agent ye-agent-i-agent yokuphatha yonke into ephantsi. I-2025, i-Ramp ithunyelwe kwiimpawu ze-500 +, yafikelela kwingeniso ye-1B ye-$, kwaye yenza konke nge-25 PMs. Benza oku ngokufuna umsebenzi ngamnye (eng, umkhiqizo, uyilo, ukuthengisa, ukuthengisa, ukuthengisa, umthetho, imali) ukuya kwibhodi kunye nokusebenza kunye nee-agent.Geoff (i-CPO Ramp) yabelane ngesakhelo sokuvavanya i-AI ifumana ubuchule obusebenzayo kakhulu: Aba bantu basenokungabikho kwinkampani ixesha elide. Ukuba awuyena umntu oqalayo kunye nengqondo yokukhula malunga nezixhobo ze-AI, uGeoff uthi kuya kuba nzima kakhulu ukukuqeqesha ukuba ugqwese. Aba bantu bazama i-AI kodwa abakawenzi umsebenzi wokwenyani ozenzekelayo.L2: Yakha i-app eyenza inxalenye yaboumsebenzi. Aba bantu banokwenza ikhowudi okanye banike impendulo enentsingiselo kumsebenzi wabanye abantu usebenzisa izixhobo ze-AI.L3: Abakhi beenkqubo. Aba bantu bakha iziseko ze-AI kunye nezakhono ezikhawulezisa wonke umntu kwiqela.Injongo yenkampani kukutyhala wonke umntu kwi-ladder. IiL0s uzikhethele ngaphandle. iiL1s ziba ziiL2. iiL2s ziba ziiL3s. Kwaye ii-L3 zinefuthe kulo lonke umbutho. UGeoff ukwabelane ngamanyathelo ama-5 anokuthi nayiphi na inkampani iwathathe ukuze abe yi-AI: Susa konke ukuxabana. Nika ukufikelela kwizixhobo ezidumileyo ze-AI ngaphandle kwemiqobo kwiithokheni okanye uhlahlo lwabiwo-mali kwaye wenze indawo yokugcina yangaphakathi yezakhono ze-AI nabani na onokutsala kuyo. Ukuba ukuseta kunzima, abantu abaninzi abayi kuyamkela.Yenza ukuthathwa komntwana kubonakale. Yenza amajelo oluntu eSlack apho abantu banokwabelana ngezinto abazenzileyo. Kuzo zonke izandla, bonisa iinjineli ezingezizo ezenza izinto ezincomekayo, njengokwakhiwa kwezemali kwi-treasury system okanye ukuthengisa i-automating website creation.Nika inkxaso yezandla. Ukusingatha iiyure zeofisi ukuba nabani na unokujoyina ukwakha izakhono ze-AI kunye nokuhamba komsebenzi. Yiba neengcali ezichongiweyo ze-AI umsebenzi wazo wonke ikukushumayela, ukumisela abantu, kwaye ubancede bafikelele “kumzuzu aha.” Landela usetyenziso kwaye ungenelele. I-Ramp ilandelela ukusetyenziswa kwethokheni kuzo zonke izixhobo ze-AI ngomsebenzi ngamnye. Ubunkokeli bubelana ngale datha ukudala uxanduva lwendalo kunye nokunyathela xa ukusetyenziswa komntu kuphantsi.Yenze ibe yimfuneko yokuqesha. Udliwano-ndlebe lwe-PM ngoku lubandakanya iseshoni ezinikeleyo apho kufuneka wakhe imveliso esebenzayo kwaye uchaze ukuba kutheni uyakhe kwaye isebenza njani. UGeoff washwankathela ifilosofi yakhe yobunkokeli kuyo yonke indima eRamp kumgca omnye: Umsebenzi wakho kukwenza umsebenzi wakho ngokuzenzekelayo. Ukusuka kuGeoff: "Ukuba ndixelele iqela lam amaxesha ali-10 ukuba i-CTA kufuneka ibe ngaphezu kwe-fold, ukulungiswa ayithethi ixesha le-11. I-encoding loo mpendulo kwi-automated design crit process okanye i-AI iphinda yenzeke loo mini. I-Ramp ibonisa indlela iinkampani ezamkela ngayo ii-arhente ze-AI, iFactory ibonisa okwenzekayo xa usakha ngeenxa zonke kuzo ukususela kusuku lokuqala. Umzi-mveliso yinkampani yophuhliso lwesoftware ye-AI yabantu abangama-55 exabisa i-300M yeedola eyakhiwe malunga ne-AI ukusuka phantsi ukuya phezulu. Nantsi into ebenza bahluke:Iinjineli zemveliso yokuQeshwa kweFactory ayiqeshi ii-PM kunye neenjineli ngokwahlukeneyo. Endaweni yoko, baqesha iinjineli zemveliso ezilawula kwaye zisebenze kunye neearhente ze-AI. Usuku oluqhelekileyo lujongeka ngolu hlobo:Jonga umkhondo osuka kwi-arhente ebalekayo ukubona apho inkqubo yenze izigqibo ezingezizo. Bhala izilungiso kungekhona njengekhowudi, kodwa njengolawulo (umzekelo, uhlaziyo kwisakhono, umgaqo omtsha we-lint, okanye i-automation ecociweyo)Hlaziya kuphela ii-PRs ukuba ii-arhente iflegi njengomngcipheko ophezulu (ii-arhente zinyamekela ukuphumla).Phakamisa imibono emitsha kunye nokubekwa phambili kweengxoxo kunye noogxa kunye ne-AI.Lo msebenzi awulula kwaye ufuna ubuchwephesha obunzulu, kodwa i-enorbase i-agent-ready agent-agent-ready agent-agent-agent agent-agent-agent agent-agent agent-agent-agent-agent-agent-agent-agent-agent-agent-agent-agent-agent-agent-agent-agent-agent-agent-agent-leve-agent-agent. codebase banokusebenza kuyo ukuze isebenze. I-Factory scores codebases kuwo wonke amanqanaba amahlanu okuvuthwa, kunye neNqanaba lesi-3 ("Standardized") kulapho uninzi lwamaqela lufuna ukujolisa kuqala.Isikhokelo sokulungela iarhente yeFactoryFactory's readiness in reusable skillsOnce your codebase is a agent-ready, inyathelo elilandelayo linika ii-arhente ulwazi lokwenza izigqibo ezilungileyo ngokusebenzisa izakhono (isiseko nje iifayile zokumakisha umbhalo). Umzi-mveliso usebenzisa izakhono ukubethelela ingcali kunye nolwazi lwenkampani kwinto enokuthi isetyenziswe nayiphi na iarhente okanye umsebenzi. Nalu uluhlu lwezakhono ezisetyenziswa yiFactory ngaphakathi kwaye ikhonkco kwiifayile zabo zokumakisha ukuze ukope kwaye uguqule: Isakhono sokulawula imveliso. Imigaqo yemveliso, isikhokelo samava eenkwenkwezi ezili-11 (ebolekwe kuBrian Chesky ka-Airbnb), itemplate ye-PRD, irubrikhi yamanqaku, kunye nesikhokelo solwimi zonke kwifayile enye yokuphawula.Isakhono sokuhlanganisa i-UI ephambili. Uyala i-Droid malunga nendlela yokwakha iimpawu usebenzisa inkqubo yoyilo lweFactory, imimiselo yomzila, kunye nemigangatho yokuvavanya.Ikhono lomhlalutyi wedatha ye-AI. Qhuba uhlalutyo lokuhlola, wakhe ukubonwa, kwaye uvelise iingxelo zezibalo usebenzisa i-Python ecosystem epheleleyo.Izakhono zezixhobo zangaphakathi. Yakha iiphaneli zolawulo, i-consoles yenkxaso, kunye needashbhodi zokusebenza ezinolawulo olululo lokufikelela kunye nokungena kuphicotho olubhakiweyo kwisakhono sekhowudi yeVibe. Ukuchwetheza ngokukhawuleza usetyenziso olutsha lwewebhu ukusuka ekuqaleni kunye nezakhelo zale mihla.Ukuba ukwazi ukufaka iikhowudi oko abantu bakho babaziyo kwizakhono, akukho mfuneko yokuba uqeshe iingcaphephe kuwo wonke umsebenzi. Amanyathelo ama-6 okukwenza konke oku ukuqheliselaUkuphinda: Ukukhwela kunye nokulawula ii-arhente ziba ngowona msebenzi uphambili kuwo wonke umsebenzi. Nazi izinto ezintandathu onokuzisebenzisa ngoku:UkusukaUmgca:Okuhlala kukho wonke umphuhlisi kwisixhobo sokukhowuda se-arhente esinje nge-Cursor, iKhowudi kaClaude, okanye iCodex. Fumana ii-PM kunye nabayili kwi-codebase. Bavumele bangenise ii-PRs kunye nekhowudi yokuthumela ngenqanawa usebenzisa ii-arhente. Yeka ukukhokela lonke utshintsho oluncinci usebenzisa injineli.Ukusuka kwiRamp:Linganisa ubuchule be-AI kwiqela lakho lonke. Isakhelo samanqanaba ama-4 eRamp sikunika isigama ekwabelwana ngaso malunga nokuba abantu baphi kwaye kufuneka baye phi. Landela usetyenziso lwe-AI kwaye uyenze ibe yinxalenye yolindelo lokusebenza. Awunako ukuphucula into ongayilinganisiyo, kwaye inkuthazo ibalulekile. Ukusuka kwiFactory:Fakela amanqaku akho ekhowudi yokulungela iarhente. Sebenzisa amanqanaba okulungela iarhente yeFactory ukuze uqonde ukuba icodebase yakho ilungile.Codify ubuchule beqela lakho kwizakhono ezinokuphinda zisetyenziswe. Faka ikhowudi eyona nto iyaziwa ngabantu bakho kwiifayile zezakhono kwaye wenze kube lula kubantu kunye neearhente ukuba bazisebenzise. Ngaphezu kwako konke, iiarhente ezikhwelayo njengokuba ungakhwela emntwini. Banike umxholo, baqhagamshele kwisitaki sakho sokusebenza, kwaye ugcine umntu enoxanduva ngeziphumo zabo.Ndazise ukuba ucinga ntoni kwizimvo!

You May Also Like

13 Trending Songs on TikTok in March 2026 (+ How to Use Them)

China’s Cybersecurity Agency Issues Security Warning About OpenClaw

Our Best AI PM Workflows for Prototyping, Strategy, and Personal OS (2026)