Umsebenzi Wakho Omusha Owoku-Inboard AI Agents: Indlela Izinkampani Zomdabu ze-AI Ezisebenza Ngayo Empeleni

In This Article

Babhalisile abathandekayo, ngichithe izinyanga ezimbalwa ezedlule ngixoxisana nabaholi ezinkampanini zomdabu ze-AI. Manje nginesiqiniseko sokuthi:Ukugibela nokuphatha ama-ejenti e-AI KUNGUmsebenzi, kungakhathaliseki ukuthi uyini umsebenzi wakho.Kulokhu kutshuza okujulile kwamahhala, ngifuna ukwabelana ngendlela izinkampani ezintathu zomdabu we-AI—Linear, Ramp, and Factory — zisebenzisa kanjani lo mgomo. Nazi ezinye izingcaphuno ezivela kumuntu ngamunye:u-Nan Yu (Inhloko Yomkhiqizo, Umugqa): “Uzoba namalungu eqembu le-AI ongawabela imisebenzi futhi ukhulume nawo ngendlela okhuluma ngayo nabantu.” UGeoff Charles (CPO, Ramp): “Uma ungasebenzisi Ikhodi ka-Claude, kungakhathaliseki ukuthi indima yakho iyini, cishe awusebenzi kahle.” U-Eno Reyes Uphethe Umkhiqizo, I-CTO, Abaphathi bedatha: ukuhlaziya, nokunye okwengeziwe ngamakhono asebenziseka kabusha angasetshenziswa yinoma yimuphi umsebenzi.” Qhubeka ufunda ukuze uthole ukubheka kwangaphakathi ukuthi inkampani ngayinye yomdabu we-AI isebenza kanjani.Linear: Ukwenza amanxusa e-AI abeqembu lesigaba sokuqalaUngakwazi @ukusho i-ejenti ukuze udale futhi unikeze inkinga enye i-ejentiIndlela ye-Linear kuma-ejenti ilolongwa umkhiqizo wabo. UNan (Inhloko Yomkhiqizo ka-Linear) ukholelwa ukuthi:Ama-ejenti kufanele abe izisebenzi ezisezingeni lokuqala. Kufanele ukwazi ukuwengeza kumaphrojekthi, uwabele ezinkingeni, futhi usho emazwaneni. Nokho, uNan uphinde akholelwe ukuthi umuntu uhlala eziphendulela ngomphumela. Nansi indlela i-Linear eyakha ngayo imikhiqizo ngama-ejenti kuso sonke isinyathelo: Qonda inkinga. Ama-ejenti afunda futhi afingqe zonke izingxoxo zamakhasimende ezivela ku-Intercom, Zendesk, ne-Gong. Badala ngokuzenzakalela izinkinga, bazikhiphe ngokumelene nokusalela emuva, futhi bazinikeze ithimba elilungile. Thola ikhambi. Njengoba abenzeli bekwazi ukufinyelela ezingxoxweni zamakhasimende, bangakusiza ukuthi uphindaphinde into ethile ngokudonsa imininingwane esekelwe idatha eziteshini eziningi. Yenza uhlelo. Ama-ejenti angahlukanisa i-spec yakho ibe amathikithi aphathekayo futhi awahambise emaqenjini alungile ngokuzenzakalelayo. Kwa-Linear, abenzeli manje bakha iningi lamathikithi. Umsebenzi womuntu ukuhlola umsebenzi wabo nokulungisa umongo ngokuhamba kwesikhathi.Kwenza. Iziphazamisi nezici ezincane zinikezwa ngokuqondile kuma-ejenti afana ne-Codex ne-Cursor. Ezicini eziyinkimbinkimbi, onjiniyela bethula Ikhodi ye-Claude futhi basebenzise i-Linear MCP ukuze bakhiphe umongo wenkinga ephelele.Kusuka ku-Nan: Kuzwakala sengathi ububanzi balokho ama-agent angakwazi ukukusingatha buyanda njalo ngekota. Amamodeli amasha namahhanisi asunduza umngcele ukusuka ekulungiseni okulula ukuya kumaphrojekthi ayanda ukuba nzima.Ufuna ukwakha ngama-ejenti afana ne-Linear? Nazi izinyathelo ezisebenzayo ezingu-4 uNan abelane ngazo ngokuthi ithimba lakho lingenzani namuhla:Wonke umthuthukisi kufanele asebenzise ithuluzi elihamba phambili le-ejenti lekhodi. Lesi isinyathelo sokuqala esilula kakhulu. Nikeza ngethuluzi elisemthethweni (Ikhesa, Ikhodi ye-Claude, noma i-Codex) futhi uyilawule ukuze ubone ukusetshenziswa.Yengeza nge-ejenti ye-async yekhodi yamafu. Abenzeli bangemuva be-Async bangakwazi ukuthwebula izinguquko ezincane kanye nokulungiswa kweziphazamisi. I-Cursor kanye no-Devin banokunikela okuhle lapha.Phikelela ukuthi abaklami nama-PM asebenze ngokuqondile ku-codebase. Ama-ejenti afana noClaude avula indlela yokungqubuzana okuphansi ukuze ama-PM nabaklami benze izinguquko ngokuqondile ku-codebase. Wonke umuntu kufanele alwele ukuba umakhi.Ama-PM nabamakethi kufanele basebenzise isixhumi esibonakalayo se-AI. Le misebenzi kufanele ibe yenza u-80-100% womsebenzi wayo ngokusebenzisa isixhumi esibonakalayo sengxoxo - kungakhathaliseki ukuthi i-Claude, i-ChatGPT, i-Notion AI, noma into efanayo. U-Nan ubona ikusasa lapho abantu bezosebenzisana nama-ejenti ezingeni elithile - echaza lokho okudingeka kwakhiwe nokuthi kungani - bese edlulisela izinto kuma-ejenti ukuze aphathe yonke into ezansi nomfula.I-Ramp: Ukulinganisa ubungcweti be-AI ngamaleveli angu-4I-Ramp iphusha ngokushesha abasebenzi babo ukuthi babe bomdabu we-AI emazingeni angu-4Uma u-Linear ebonisa indlela yokwenza ama-ejenti abe yisigaba sokuqala senkampani yakho bese uthola ukuthi ingxenye yeqembu lakho lonke i-Ramp ikhombisa kanjani. Ngo-2025, i-Ramp yathumela izici ezingu-500+, yafinyelela ku-$1B engenayo, futhi yakwenza konke ngo-25 PMs. Benza lokhu ngokudinga umsebenzi ngamunye (eng, umkhiqizo, idizayini, ukuthengisa, ukumaketha, okusemthethweni, ezezimali) ukuba bagibele futhi basebenze nama-ejenti.UGeoff (I-CPO Ramp) wabelane ngohlaka lokuhlola ubuchwepheshe be-AI obuthola busebenziseka kakhulu: I-ChatL Laba bantu cishe ngeke babe senkampanini isikhathi eside. Uma ungeyena umuntu oziqalayo onomqondo wokukhula obheke kumathuluzi e-AI, u-Geoff uthi kuzoba nzima kakhulu ukukuqeqesha ukuze uphumelele.L1: Isebenzisa futhi ilungise ama-GPT, amaphrojekthi, namathuluzi angaphakathi e-AI. Laba bantu bazama i-AI kodwa abakawenzi umsebenzi wangempela ozenzakalelayo. L2: Yakha uhlelo lokusebenza olusebenzisa ingxenye yabo ngokuzenzakalelayo.umsebenzi. Laba bantu bangenza ikhodi noma banikeze impendulo enengqondo ngomsebenzi wabanye abantu besebenzisa amathuluzi e-AI.L3: Abakhi bezinhlelo. Laba bantu bakha ingqalasizinda ye-AI kanye namakhono asheshisa wonke umuntu eqenjini.Inhloso yenkampani ukuphusha wonke umuntu esitebhisini. Ama-L0 azikhethele wona. Ama-L1 abe ama-L2. Ama-L2 abe ama-L3. Futhi ama-L3 athonya yonke inhlangano. U-Geoff uphinde wabelane ngezinyathelo ezi-5 noma iyiphi inkampani engazithatha ukuze ibe yi-AI:Susa konke ukungqubuzana. Nikeza ukufinyelela kumathuluzi e-AI adumile ngaphandle kwemingcele kumathokheni noma kwisabelomali futhi udale inqolobane yangaphakathi yamakhono e-AI noma ubani angadonsa kuwo. Uma ukusetha kunzima, abantu abaningi ngeke bakuthathe.Yenza ukutholwa kubonakale. Dala amashaneli e-Slack asesidlangalaleni lapho abantu bengabelana ngalokho abakwakhile. Ngaso sonke isikhathi, bonisa okungebona onjiniyela abenza izinto ezihlaba umxhwele, njengokwakha ngezezimali isistimu ye-treasury noma ukwakhiwa kwewebhusayithi ezenzakalayo zokumaketha.Nikeza usekelo olusebenzayo. Amahora okusingatha amahhovisi noma ubani angawajoyina ukuze akhe amakhono e-AI nokugeleza komsebenzi. Yiba nochwepheshe abaqokiwe be-AI abamsebenzi wabo wonke kuwukushumayela ivangeli, ukumisa abantu, nokubasiza bafinyelele “esikhathini esimnandi.” Landelela ukusetshenziswa futhi ungenelele. Irempu ilandelela ukusetshenziswa kwamathokheni kuwo wonke amathuluzi e-AI ngomsebenzi ngamunye. Ubuholi babelana ngale datha ukuze budale ukuziphendulela okungokwemvelo futhi bungenele lapho ukusetshenziswa kothile kuphansi. Kwenze kube imfuneko yokuqasha. Izingxoxo ze-PM manje zifaka iseshini ezinikele lapho udinga ukwakha umkhiqizo osebenzayo bese uchaza ukuthi kungani uwakhile nokuthi usebenza kanjani. U-Geoff wafingqa ifilosofi yakhe yobuholi kuyo yonke indima e-Ramp emgqeni owodwa:Umsebenzi wakho ukwenza umsebenzi wakho ngokuzenzakalelayo.Kusuka ku-Geoff: "Uma ngitshela ithimba lami izikhathi ezingu-10 ukuthi i-CTA idinga ukuba ngaphezu kwe-fold, ukulungisa akusho okwesi-11. Kufaka ikhodi leyo mpendulo ku-automated design crit processy noma usuku olulodwa lwe-AI lungaphinda lwenzeke lelo khono le-AI." I-Ramp ikhombisa ukuthi izinkampani zamukela kanjani ama-ejenti e-AI, Imboni ikhombisa ukuthi kwenzekani uma wakha eduze kwabo kusukela ngosuku lokuqala. I-Factory yinkampani yokuthuthukisa isoftware ye-AI yabantu abangama-55 enenani lama- $ 300M eyakhiwe yazungeza i-AI kusukela phansi kuya phezulu. Nakhu okubenza bahluke:Ifektri yokuqasha ayiqashi ama-PM nonjiniyela ngokuhlukana. Esikhundleni salokho, baqasha onjiniyela bemikhiqizo abaphatha futhi basebenze nabasebenzeli be-AI. Usuku olujwayelekile lubukeka kanjena: Hlola ukulandelelwa kokusebenza komenzeli ukuze ubone lapho isistimu yenze khona izinqumo ezimbi. Bhala ukulungisa hhayi njengekhodi, kodwa njengokubusa (isb., isibuyekezo sekhono, umthetho omusha we-lint, noma umshini ozenzakalelayo ocwengisisiwe)Buyekeza kuphela ama-PRs ama-ejenti ahlaba umkhosi njenganobungozi obukhulu (ama-ejenti anakekela okunye).Phakamisa imibono emisha kanye nokubeka kuqala impikiswano nozakwenu kanye ne-AI.Lo msebenzi awulula futhi udinga ubuchwepheshe obujulile, kodwa isidingo esikhulu sokufunda ikhodi yisilinganiso esikhulu sokufunda kwakho. codebase angasebenza kuwo ukuze asebenze ngempumelelo. Ifekthri ithola izisekelo zekhodi kumaleveli okuvuthwa amahlanu, futhi Izinga 3 ("Okujwayelekile") yilapho amaqembu amaningi edinga ukuhlose khona kuqala.Uhlaka lokulungela umenzeli wefekthriFaka ulwazi ube ngamakhono asebenzisekayoUma i-codebase yakho isilungele i-ejenti, isinyathelo esilandelayo sinikeza ama-ejenti ulwazi lokwenza izinqumo ezinhle ngamakhono (okuyisisekelo nje bhala amafayela wokumaka). I-Factory isebenzisa amakhono ukuze ifake ulwazi lwengcweti nolwazi lwenkampani kokuthile okungase kusetshenziswe noma yimuphi umenzeli noma umsebenzi. Nalu uhlu lwamakhono i-Factory ewasebenzisayo ngaphakathi kanye nezixhumanisi kumafayela ayo okumaka ukuze uwakopishe futhi uwalungise:Ikhono lokuphatha umkhiqizo. Izimiso zomkhiqizo, uhlaka lwesipiliyoni sezinkanyezi ezingu-11 (esibolekwe ku-Brian Chesky we-Airbnb), isifanekiso se-PRD, irubrikhi yamaphuzu, nesiqondiso solimi konke kufayela elilodwa lokumaka.Ikhono lokuhlanganisa i-UI yangaphambili. Iyalela i-Droid ukuthi ingazakha kanjani izici kusetshenziswa isistimu yokuklama Yefekthri, izimiso zomzila, namazinga okuhlola.Ikhono lokuhlaziya idatha ye-AI. Qalisa ukuhlaziya okuhlola, yakha okubonwayo, futhi ukhiqize imibiko yezibalo usebenzisa i-ecosystem ephelele ye-Python.Ikhono lamathuluzi angaphakathi. Yakha amaphaneli okuphatha, ama-consoles asekelayo, namadeshibhodi asebenzayo anezilawuli ezifanele zokufinyelela kanye nokungena okubhaliwe okubhakwe ngekhono lokubhala nge-Vibe. Yenza ngokwezifiso izinhlelo zokusebenza ezintsha zewebhu kusukela ekuqaleni ngezinhlaka zesimanje.Uma ukwazi ukufaka amakhono abantu bakho abangcono kakhulu abakwaziyo, awudingi ukuqasha ochwepheshe kuwo wonke umsebenzi. Izinyathelo ezi-6 zokusebenzisa konke lokhuUkuphindaphinda: Ama-ejenti okugibela nawo azoba umsebenzi oyinhloko wawo wonke umsebenzi. Nazi izinto eziyisithupha ongazisebenzisa njengamanje:KusukaI-Linear:Okuzenzakalelayo wonke unjiniyela ethuluzini le-ejenti lekhodi njenge-Cursor, Claude Code, noma i-Codex. Thola ama-PM nabaqambi ku-codebase. Bavumele bahambise ama-PRs futhi bathumele ikhodi besebenzisa ama-ejenti. Yeka ukuthungatha lonke ushintsho oluncane ngonjiniyela.Kusuka ku-Ramp:Linganisa ubuhlakani be-AI eqenjini lakho lonke. Uhlaka lwamaleveli angu-4 lwe-Ramp lukunikeza ulwazimagama olwabiwe lokuthi abantu bakuphi nokuthi badinga ukuya kuphi. Landelela ukusetshenziswa kwe-AI futhi ukwenze kube yingxenye yokulindelwe kokusebenza. Awukwazi ukuthuthukisa lokho ongakukali, futhi izikhuthazo zibalulekile. Kusuka Embonini:Faka i-codebase yakho ukuze ulungele umenzeli. Sebenzisa amazinga okulungela umenzeli Wefekthri ukuze uqonde ukuthi i-codebase yakho isilungile yini.Hlela ubuhlakani beqembu lakho libe ngamakhono angasetshenziswa kabusha. Faka ikhodi lokho abantu bakho abangcono kakhulu abakwaziyo kumafayela amakhono futhi wenze kube lula ngabantu nama-ejenti ukuthi bawasebenzise. Ngaphezu kwakho konke, ama-agent angaphakathi afana nokuthi ungagibela umuntu. Banikeze umongo, baxhume esitaki sakho sokusebenza, futhi ugcine umuntu enesibopho ngemiphumela yabo.Ngazise ukuthi ucabangani kumazwana!

You May Also Like

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

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

China’s Cybersecurity Agency Issues Security Warning About OpenClaw