Ama-GPT enziwe ngokwezifiso ayavalwa. Okudingeka ukwazi.

Impendulo emfushane: I-OpenAI iqeda ama-GPT enziwe ngokwezifiso: izinhlelo zomuntu siqu azikwazi ukudala noma ukushicilela ezintsha, futhi ama-bot akhona awusizo lwesikhashana nje. Uma i-GPT isathola imali yayo, thumela imiyalelo namafayela olwazi manje, bese wakha kabusha kumaphrojekthi, ama-plugin, noma ama-Workspace Agents.

Izinto ezibalulekile okufanele uzicabangele:

Azikho izakhiwo ezintsha: Yeka ukudala ama-GPT enziwe ngokwezifiso ezinhlelweni zomuntu siqu; phatha akhona njengesikhashana.

Isitoko kuqala: Bhala uhlu lwezinhloso ze-GPT ngayinye, umnikazi, abantu abaxhomeke kuyo, amafayela, kanye nezinto eziza kuqala ngaphambi kokuthuthela kwenye indawo.

Thumela ubuchopho: Kopisha imiyalelo ephelele bese ulanda wonke amafayela olwazi ngenkathi usakwazi.

Yakha kabusha ngokujulile: Thutha amahhashi asebenzayo aphezulu abe yi-ChatGPT Projects; susa amathoyizi ngaphandle komcimbi.

Xwayisa abahlanganyeli: Tshela amaqembu nabasebenzisi bezixhumanisi ezabiwe ngaphambi kokuba izixhumanisi zibe mnyama futhi imisebenzi iphuke.

Kuyini ukuphela uma kuqhathaniswa nalokho okusasebenza njengamanje 

Ukudalwa nokushicilelwa kwama-GPT amasha ngokwezifiso ezinhlelweni zomuntu siqu sekuvele kuvaliwe. Awukwazi ukuvula uhlu olusha lwesitolo se-GPT noma ukushicilela i-GPT entsha ngokwezifiso ngendlela obuyenza ngayo. Ekhona ingase isasebenziseka - futhi kwezinye izimo ingahlelwa - okwesikhashana. Phatha "kusasebenza" njengomusa wesikhashana, hhayi isithembiso.

Ukusethwa kwezindawo zokusebenza kanye nezinkampani kulandela indlela yokuphela eshicilelwe. Lokho ngokuvamile kusho izaziso zokuphatha, amathuluzi okufuduka, kanye nokusunduza amabhlogo okwakha amasha: Amaphrojekthi e-ChatGPT ezindawo zokusebenza ezigxile kumafayela nemiyalelo, ama-plugin lapho etholakala khona, kanye nama-Workspace Agents okwenzakalela kweqembu. Amagama ayashintsha kancane ngohlelo; umqondo awushintshi. Ama-GPT enziwe ngokwezifiso njengobuso bomkhiqizo asezophela.

Yini esasebenza ngokoqobo, ngenkathi usakwazi:

• Ukuvula nokuxoxa nama-GPT amaningi akhona angokwezifiso osuvele unawo noma owabelwana ngawo
• Ukuhlela imiyalelo noma amafayela olwazi kwezinye izinhlelo ngaphambi kokuba ukufinyelela kokuhlela kunyamalale
• Ukukopisha imiyalelo yesistimu, iziqalisi zengxoxo, kanye namafayela olwazi alayishiwe
• Ukubhala phansi ukuthi ubani othembele ku-GPT ngaphambi kokuba wonke umuntu athole indlela enzima

Okungasasebenzi kuma-akhawunti amaningi omuntu siqu: ukudala ama-GPT amasha enziwe ngokwezifiso, ukushicilela esitolo se-GPT, nokuphatha ama-GPT enziwe ngokwezifiso njengekhaya lesikhathi eside lemisebenzi ebalulekile. Uma ama-GPT enziwe ngokwezifiso evalwa. Okudingeka ukwazi kuqala lapha - yeka ukwakha amasha bese uqala ukulondoloza amahle.

Indawo yokusebenza yomuntu siqu vs indawo yokusebenza: kungani uhlelo lwakho lushintsha incwadi yokudlala

Abasebenzisi be-Personal Free / Go / Plus / Pro sebevele basendleleni "engenazo izakhiwo ezintsha". Umsebenzi wakho ukwenza isipele somuntu siqu: thatha imiyalelo, thatha amafayela, wakhe kabusha ngaphakathi kwe-Projects noma elinye ithuluzi, bese utshela noma ubani owabelane naye nge-GPT ukuthi isixhumanisi singase sibe mnyama.

Abaphathi bebhizinisi kanye nabaphathi bendawo yokusebenza babhekene nobunzima obuhlukile - futhi, ngokumangalisayo, imephu ecacile. Ama-GPT abiwe avame ukuhlala phakathi kwenqubo yeqembu: ama-robot angena ngemvume, imibuzo nezimpendulo zenqubomgomo, amathrekhi engxoxo yokuthengisa, ama-macro asekelayo agqoke njengengxoxo. Lapho lawo ephuka, abantu abawaniki amahlombe; amathikithi ayavela. Ngakho-ke beka uhlu lwezinto ezikhona ezingeni lenhlangano, wabele abanikazi, bese ukhetha abazolandela (Amaphrojekthi, ama-plugin, ama-Workspace Agents) ngaphambi kokuba lawo abiwe ayeke ukusebenza.

Ngicabanga ukuthi umthetho oyinhloko uthi: uma uyisebenzisa kuphela, yisekele kuleli sonto. Uma iqembu liyisebenzisa, yiphathe njengokungasasebenzi komkhiqizo ngohlelo oluncane lokufuduka. Akuyona i-PRD enamakhasi angu-40. Uhlu lokuhlola kanye nomnikazi oqanjwe ngegama. Lokho kwanele.

Uhlu lokuhlola uhlu lwempahla: thola yonke i-GPT yangokwezifiso ebalulekile

Awukwazi ukuthutha lokho ongakwazi ukukusho. Vula uhlu lwakho lwe-GPT bese wenza uhlu olulula ngaphambi kwanoma yini enye. Ispredishithi, uhlelo lokusebenza lwamanothi, amanothi anamathelayo - noma yini. Ukuphelela kuyisitha lapha.

Ku-GPT ngayinye eyenziwe ngokwezifiso, thatha:

• Igama kanye nenhloso - umusho owodwa ngalokho elikwenzayo
• Umnikazi - wena, umuntu osebenza naye eqenjini, noma "oyintandane ngandlela thile"
• Ubani oncike kukho - wena, ithimba, amaklayenti, noma isixhumanisi esabiwe esizungeza iSlack
• Ubude bemiyalelo - isifinyezo esifushane vs inoveli yemithetho
• Amafayela olwazi - ama-PDF, amadokhumenti, ama-CSV alayishwe kuwo
• Izenzo / amathuluzi - noma yiziphi izikhonkwane ze-API, ukuphequlula, noma izenzo ezenziwe ngokwezifiso
• Okubalulekile - okubalulekile kwansuku zonke, okuhle kwamasonto onke, noma uthuli lwedijithali
• Ukuqagela komlandeli - Iphrojekthi, i-plugin, i-Workspace Agent, noma "ukuthatha umhlalaphansi buthule"

Hlela ngokubaluleka kuqala, hhayi ngokuthi isithombe se-GPT besisihle kangakanani. Isithombe ngeke sikuphuthele. Umsebenzi wakho wangoMsombuluko ekuseni ungase ube ngcono.

Abantu abaningi bathola, cishe ngengozi, ukuthi bane-GPT ezintathu ezenza umsebenzi ofanayo ngemiyalelo ehlukene kancane. Hlanganisa ngenkathi ufuduka. Ikusasa - uzothumela incwadi yokubonga.

Okufanele ukonge, kungani, kanjani, nokuthi kuphuthuma kangakanani

Sebenzisa lokhu njengohlu lokupakisha. Kufakwe izinto ezincane ezingavamile - yebo, "iziqalo zengxoxo" zifanele ukugcinwa ngisho noma ungaphinde uzibonise kubasebenzisi.

Yenza imigqa ebalulekile kuqala. Amafayela olwazi ikakhulukazi - uma ukufinyelela kokuhlela noma ukulanda sekuthambile, ukufuna i-PDF yokuqala kwi-laptop yomunye umuntu kuyinhlupho ekhethekile.

Thumela imiyalelo namafayela olwazi ngaphambi kokuba ayeke ukusebenza

Lena yingxenye engathandeki ekusindisa kamuva. Vula i-GPT ngayinye ebaluleke kakhulu ngesikhathi usakwazi. Kopisha imiyalelo ephelele kumbhalo ocacile noma ifayela le-markdown - linda, umbhalo ocacile uphephile ekulawulweni kwenguqulo; i-markdown nayo ilungile. Qamba amafayela ngokucacile: support-triage-instructions.txt, hhayi i-final-final-v3.txt.

Bese ulanda ifayela lolwazi ngalinye. Lifake kufolda ngayinye ye-GPT. Uma ifayela libuyekeziwe ngaphakathi kwe-GPT futhi ungaqiniseki ukuthi iyiphi inguqulo yamanje, thatha okuku-GPT manje bese ubheka umthombo weqiniso ku-Drive noma ku-SharePoint. Ukusekela kabili kudlula i-PDF eyodwa eyimfihlakalo.

Amathiphu ambalwa awusizo azwakala ecacile uze uweqe:

• Thumela ngosuku oluzolile, hhayi imizuzu emihlanu ngaphambi kombukiso
• Susa izimfihlo embhalweni wemiyalelo uma uwunamathisela kudrayivu yethimba elabiwe
• Qaphela izici zemodeli othembele kuzo - imithetho yethoni, imithetho yokufometha, "ungalokothi usungule ama-SKU," into enjalo
• Uma i-GPT isebenzisa izenzo ezenziwe ngokwezifiso, bhala phansi ukuthi isenzo ngasinye senzeni ngolimi lwabantu, hhayi nje uhlelo

Ngiqale lesi sigaba ngiqonde ukuthi "vele ukopishe umyalezo," ngabe sengibona ukuthi ingxenye yenani ivame ukungcwatshwa kumafayela. Isaziso yikhadi leresiphi; amafayela olwazi ayi-pantry. Udinga kokubili noma isidlo sakusihlwa siyi-toast. Ngokwesingathekiso. Sicela ungadli ama-PDF akho.

Yakha kabusha ama-robot abalulekile kumaphrojekthi e-ChatGPT 🛠️

Kuma-workflow amaningi omuntu siqu kanye neqembu, ama-ChatGPT Projects ayikhaya eliseduze elingokwemvelo. I-Project ikunikeza indawo ezinikele enemiyalelo namafayela ayo - eduze ngokwanele ne-GPT yangokwezifiso kangangokuthi ukufuduka kuzwakala kujwayelekile, ngaphandle kokwenza sengathi kuyi-clone yokuchofoza okukodwa.

Indlela yokwakha kabusha enengqondo:

• Dala iphrojekthi eqanjwe ngomsebenzi we-GPT endala, hhayi igama lehlaya le-GPT endala
• Namathisela imiyalelo ethunyelwe; nquma noma yini enengqondo kuphela esitolo se-GPT
• Layisha amafayela olwazi owadingayo ngempela - ungalahli yonke imibhalo engenalutho ehleli kufolda
• Hlola ngemiyalelo efanayo owayisebenzisa ngaphambili bese uqhathanisa imiphumela eceleni
• Mema abantu abasebenzise i-GPT endala bese ususa isixhumanisi esidala uma sebekhululekile

Amaphrojekthi asebenza kahle kakhulu emsebenzini oqhubekayo: amafolda ocwaningo, amapayipi okuqukethwe, ama-client briefs, imibuzo nezimpendulo zangaphakathi ezine-corpus engaguquki. Akuwona "uhlelo lokusebenza lomphakathi esitolo se-GPT" kodwa ayizindawo zokusebenza ezingathi sína. Kulungile lokho. Iningi lama-GPT enziwe ngokwezifiso athola ukugcinwa kwawo ayengabasebenzi abasebenzayo, hhayi izitolo.

Uma ama-GPT enziwe ngokwezifiso evalwa. Okudingeka ukwazi ngokwakhiwa kabusha yilokhu: qala ngama-bots akho amathathu abalulekile, hhayi i-zoo yonke. Thumela lawo. Bese unquma ukuthi yiliphi kulawa asele elifanelwe yiProjekthi nokuthi yiliphi elifanelwe ukuvaleliswa buthule.

Ama-plugin, Ama-ejenti Endawo Yokusebenzela, kanye nezinye izindlela zokulandela 🔌

Akuzona zonke izinto okufanele zibe seProjekthini. Amanye ama-GPT enziwe ngokwezifiso ayeyizimbozo ezincane ezizungeze amathuluzi, ama-API, noma izinqubo zeqembu ezinezinyathelo eziningi. Lokho kuhambisana kangcono nama-plugin lapho uhlelo lwakho luwasekela khona, noma kuma-Workspace Agents ekusethweni kwesitayela sebhizinisi.

Ukumapha okungahleliwe okusebenze ngokoqobo:

• Imiyalelo-esindayo + amafayela - Amaphrojekthi e-ChatGPT
• Ithuluzi / ukuzenzekela kwe-API - ama-plugin noma abalandeli besitayela sesenzo; phinda uqinisekise ngokucophelela
• Izinqubo zeqembu ngobunikazi - Ama-Ejenti Endawo Yokusebenzela, nomphathi onolwazi
• Ama-bot omphakathi / okutholakele - cabanga kabusha ngokusabalalisa; isikhathi se-GPT Store sama-GPT enziwe ngokwezifiso siyaphela
• Amathoyizi akhishwa kanye kuphela - ungathuthi; gcina umyalo bese uqhubeka

Lindela ukungqubuzana okuthile. Amakhono ngeke afane nomuntu ngamunye. I-GPT eyenziwe ngokwezifiso ehlanganisa ukuphequlula, amafayela, kanye nobuntu obungavamile ingaba yiProjekthi kanye nethuluzi elihlukile, noma i-ejenti enezimvume ezicacile. Lokho kuhlukana kuyacasula futhi kunempilo - amabhokisi amnyama ambalwa ayimfihlakalo.

Ngaphandle kwe-ChatGPT ngokuphelele, amanye amaqembu akha kabusha imisebenzi efanayo kwabanye abasizi noma izinhlelo zokusebenza zangaphakathi. Lokho kuyasebenza. Gcina nje imiyalelo namafayela akho athunyelwe ukuze ungabhali kabusha kusuka enkumbulweni njengendaba yabantu.

Iqembu kanye nezingozi ze-GPT ezabiwe okufanele ungazinaki

Ama-GPT Ahlelwe Ngokwezifiso abiwe yilapho ukuvala okuthule kuba khona umsindo. Othile ubeka uphawu lwe-GPT ekhasini le-Notion elibizwa ngokuthi "Indlela esisekela ngayo." Omunye umuntu ubeka isixhumanisi sokufaka. Akekho ongumnikazi wemiyalelo. Ngemuva kwalokho ukudala kuyakhiya, ukuhlela kuyama, futhi ngokuzumayo i-bot "wonke umuntu ayisebenzisayo" iqala ukusebenza njenge-attic enezinzwa.

Nciphisa ukucindezeleka kusenesikhathi:

• Nika umnikazi nge-GPT eyabiwe - umuntu onegama, hhayi "ithimba"
• Bhala inothi elifushane eziteshini eziyisebenzisayo: yini eshintshayo, yini olandela, nini ukushintsha
• Shintsha izixhumanisi ezifakwe amakhodi kuma-wiki kanye namadokhumenti okubhalisa ngenkathi i-GPT endala isaphendula
• Gcina ingobo yomlando yemiyalelo yokufunda kuphela ukuze abasebenzi abasha bangafundi ku-URL efile
• Kuma-GPT abhekene namakhasimende, hlela umlayezo ozwakala uzolile, hhayi odabukisayo

Isikhathi sokubhuqa esincane: uma inqubo yakho ihlala ngaphakathi kwe-GPT yangokwezifiso ngaphandle kokusekelayo, lokho kwakungeyona inqubo. Lokho kwakuyithemba elinesixhumi sokuxoxa. Lungisa ithemba ngenkathi isixhumi sisavuleka.

Amaphutha avamile ngesikhathi sokufuduka (sicela uwagweme lawa)

Abantu benza amaphutha afanayo ngaphansi kwengcindezi encane. Ngiwenze ambalwa awo. Ukufunda kuyatholakala; ukuphinda kuyinto ongayikhetha.

• Silinde isimemezelo esiphelele - yenza ngesikhathi usakwazi ukuhlela nokulanda
• Ukuthutha konke - ukuthutha okubalulekile nokusetshenziswa kakhulu kuphela; gcina okusele kungobo yomlando
• Ukukhohlwa amafayela olwazi - imiyalelo engenamafayela iyingxenye yobuchopho
• Ukunamathisela izimfihlo kumadokhumenti abiwe - okhiye be-API abafanele i-wiki yeqembu
• Ukushintsha ithoni nemithetho ngesikhathi sokufuduka - yakha kabusha kuqala, uhlele kabusha okwesibili
• Ukungatsheli iqembu - ukunqamula buthule kudala amathikithi okusekela nokungathembani
• Uma sicabanga ukuthi amaphrojekthi afana - hlola imiphumela; lungisa; bese umemezela ukuthi "kwenziwe"
• Ukushiya uhlu lwezitolo ze-GPT njengomthombo weqiniso - kopisha ikhophi manje

Futhi: ungasunguli indawo yokufuduka eyinkimbinkimbi. Awudingi ikomidi eliqondisayo le-GPT ephindaphinda okuqukethwe esetshenziswa kabili ngenyanga. Udinga ifolda nemizuzu engamashumi amabili.

Uhlelo olusebenzayo olumise okwesonto ngaphandle kokwesaba kwekhalenda

Khuluma ngamagama ahlobene bese uqhubeka. Nansi indlela elula esebenza ngayo kungakhathaliseki ukuthi une-GPT eyodwa noma amashumi amabili.

1. Namuhla: Isitoko. Maka okubalulekile uma kuqhathaniswa nokuthatha umhlalaphansi.
2. Isikhathi esilandelayo samahhala: Thumela imiyalelo namafayela olwazi nganoma yini ebalulekile noma ephezulu.
3. Ngokushesha ngemva kwalokho: Yakha kabusha ezimbili noma ezintathu eziphezulu kumaProjekthi (noma ama-ejenti/ama-plugin uma lokho kufaneleka).
4. Ngaphambi kokuba bayeke ukusebenza: Shintsha okuzenzakalelayo kwakho; buyekeza izixhumanisi zeqembu; tshela abathintekayo.
5. Okuqhubekayo: Susa ama-GPT anenani eliphansi; gcina ingobo yomlando; yeka ukudala ama-GPT amasha enziwe ngokwezifiso angenakushicilelwa noma kunjalo.

Ingxenye engokomzwelo inzima kunengxenye yobuchwepheshe, kaningi kunalokho abantu abakuvumayo. Uchithe ubusuku ulungisa i-GPT ekugcineni "yathola" izwi lakho. Ukuvalelisa kuzwakala kuyisiwula uze ukhumbule ukuthi usenemiyalelo. Umphefumulo we-bot ungumbhalo. Umbhalo uyahamba.

Izinto ezibalulekile okufanele uzicabangele

Ama-GPT enziwe ngokwezifiso ayavalwa. Okudingeka ukwazi, ngamafuphi:

• Ukudalwa nokushicilela okusha kwe-GPT ngokwezifiso ezinhlelweni zomuntu siqu sekuvele kunqunyelwe; ezikhona zisesikhathini esibolekiwe
• Faka konke ohlwini; beka phambili ngokusetshenziswa okuphilayo, hhayi ukukhumbula izinto ezidlule
• Thumela imiyalelo, amafayela olwazi, iziqalo, kanye namanothi esenzo ngenkathi usakwazi
• Yakha kabusha imisebenzi ebalulekile kumaphrojekthi e-ChatGPT; sebenzisa ama-plugin noma ama-Workspace Agents lapho efanelana kangcono khona
• Xwayisa amaqembu, buyekeza amadokhumenti, nika abanikazi ama-bot abiwe
• Yeqa ukwesaba, yeqa udoti ofudukayo, yeqa ulinde isikhathi esifanele

Thatha isinyathelo kusenesikhathi, paka ngokucophelela, wakhe kabusha lokho okuyitholayo. Ipulatifomu iyashintsha ngaphansi kwe-GPT Store kanye nesendlalelo se-Custom GPT - lokho akukuhle, akupheli umsebenzi. Imisebenzi yakho yayihlala ingaphezu kwenkinobho yomkhiqizo owodwa. Gcina izindlela zokupheka. Shintsha ikhishi. Uzolunga.

Isibonelo esisebenzayo: Ukuthuthela ukwesekwa okwabiwe kwe-GPT ngokwezifiso kuphrojekthi ye-ChatGPT

Ukwehla kwentengo kuzwakala kungaqondakali kuze kube yilapho isixhumanisi sokungeniswa ngoMsombuluko sinyamalala. Nansi indlela umholi wokusekelwa kwamakhasimende wase-UK asebenzise ngayo isitokwe → ukuthekelisa → indlela yokwakha kabusha kusuka kulo mhlahlandlela lapho ama-GPT enziwe ngokwezifiso evala - nokuthi yini okudingeka ukwazi ukuze upakishe i-bot ethola imali yayo.

Isimo

Ithimba likaJordan line-Custom GPT ebizwa ngokuthi "Support Triage Helper." Ligcina imithetho yethoni, "ungalokothi usungule ama-SKU," iphakethe lolwazi lwezihloko zosizo, kanye neziqalo ezine zengxoxo lapho abaqashi abasha bechofoza ngaphandle kokucabanga. Isixhumanisi sitholakala ku-Notion ngaphansi kokuthi "Indlela esisekela ngayo." Ukudalwa kwama-Custom GPT amasha ezinhlelweni zabo zesitayela somuntu siqu sekuvele kuvaliwe. I-bot eyabiwe isavuleka - okwamanje - okuwumusa, hhayi isivumelwano sokuqasha.

UJordan akathuthi yonke i-zoo yama-GPT aqediwe uhhafu. Okubaluleke kakhulu yile ndlela yokusebenza: kopisha imiyalelo, landa amafayela olwazi, wakhe kabusha ngaphakathi kwe-ChatGPT Project, buyekeza isixhumanisi se-wiki, tshela isiteshi, bese ususa i-URL endala lapho abantu bekhululekile.

Umgomo ukuqhubeka ngaphandle kwethiyetha yokufuduka - uhlu lokuhlola kanye nomnikazi oqanjwe ngegama, hhayi i-PRD enamakhasi angu-40.

Okudingwa ukufuduka

• Umugqa wempahla: igama, injongo, umnikazi (iJordani), oncike kuyo, amafayela olwazi, izenzo, okubalulekile, ukuqagela okulandelayo (Iphrojekthi)
• Imiyalelo ephelele eyenziwe ngokwezifiso inamathiselwe kufayela eliqanjwe ngokucacile (isb. support-triage-instructions.txt)
• Lonke ifayela lolwazi elilayishiwe lilayishiwe kufolda eqanjwe nge-GPT
• Iziqalo zengxoxo kanye nezingxoxo eziyisampula ezingu-3-5 ezibonisa ukuthi "kuhle" kanjani
• Amanothi nganoma yiziphi izenzo/izikhonkwane ze-API olimini lwabantu (azikho izimfihlo ku-Slack)
• Iphrojekthi ye-ChatGPT enemiyalelo eqoshiwe, amafayela kuphela asadingeka, kanye nemiyalelo yokuhlola eceleni
• Kubuyekezwe izixhumanisi ze-Notion/onboarding kanye ne-channel note emfushane ngaphambi kokuba i-GPT endala iyeke ukusebenza

Isibonelo semiyalelo

Sebenzisa lokhu njengombhalo wokwakha kabusha ngaphakathi kwephrojekthi entsha (ngemuva kokunamathisela imiyalelo yesistimu ethunyelwe futhi ulayishe amafayela olwazi aqinisekisiwe):

UnguMsizi Wokusekela i-Triage we-SaaS yethu yokuhlela. Landela ithoni ethunyelwe ngaphandle futhi "ungalokothi usungule ama-SKU" ngqo. Phendula kuphela ezihlokweni zosizo ezilayishiwe; uma impendulo ingekho, yithi awazi bese uphakamisa indlela yokwenyuka komuntu. Umphumela: imbangela engaba khona, ubufakazi obuvela kumadokhumenti, isheke elilandelayo, kanye nempendulo ebhaliwe ngesiNgisi sase-UK. Akukho sandulelo. Uma umbuzo udinga i-API noma isenzo sesistimu esibukhoma esingakaxhumani lapha, shono esikhundleni sokwenza sengathi.

Ukuhlolwa kokuqala okufanele kwenziwe (imiyalelo efanayo ne-GPT endala): (1) "Ikhasimende lithi ukushintshashintsha kokukhokhisa kuphuke ngemva kokuthuthukiswa - triage," (2) "Bhala impendulo ezolile uma singayitholi i-invoyisi yabo," (3) isiqalisi ngasinye sengxoxo endala kanye. Qhathanisa imiphumela nezingxoxo zesampula ezigciniwe ngaphambi kokuthi umemezele ukushintsha.

Indlela yokuyihlola

• Isitokwe kuqala: bhala yonke i-GPT eyenziwe ngokwezifiso; maka okubalulekile / okuphezulu / umhlalaphansi. Ungaqali ngama-avatar amahle.
• Thumela ngenkathi ukuhlela/ukulanda kusasebenza: imiyalelo + amafayela olwazi + iziqalisi. Ama-PDF ayimfihlo anesipele esiphindwe kabili avela ku-Drive uma izinguqulo zihlukile.
• Yakha kabusha ezimbili noma ezintathu eziphezulu kuphela. Hlola izimpendulo ezifanayo eceleni kwe-GPT endala ngenkathi isasebenza.
• Icala le-Edge: i-GPT enezenzo ezenziwe ngokwezifiso - bhala phansi lokho okwenziwe yisenzo ngasinye; phinda uqinisekise ngokucophelela kuma-plugin noma kuma-ejenti uma lokho kufaneleka kangcono kuneProjekthi.
• Ukuhlolwa kokwamukelwa ngaphambi kokusika: (1) ifayela lemiyalelo liqediwe, (2) wonke amafayela olwazi olubalulekile akhona, (3) izixwayiso ezintathu zokuhlola zifanisa ikhwalithi yesampula, (4) Isixhumanisi sombono sibuyekeziwe, (5) isiteshi esinolwazi ngegama lomnikazi, (6) azikho izinkinobho ze-API ezinamathiselwe kwi-wiki yeqembu.

Umphumela

Umphumela obonisayo (isibonelo sokulinganisa kweqembu elilodwa lokusekela ukufuduka, hhayi ucwaningo lwe-OpenAI olushicilelwe): Kuzo zonke i-GPTs eziyi-8 ezenziwe ngokwezifiso ezifakwe ohlwini, ezintathu zazibalulekile/ziphezulu futhi zakhiwa kabusha njengeProjekthi; ezinhlanu zigcinwe ngemiyalelo kuphela. Ukuthumela nokwakha kabusha i-bot yokusekela triage kuthathe cishe imizuzu engama-90 iwashi lodonga (i-inventory 15, ukuthumela ngaphandle 25, ukusethwa kwephrojekthi kanye nokulayishwa kwamafayela angu-20, ukuhlolwa eceleni 30). Ngemuva kokusika, "i-GPT iye kuphi?" ama-ping esiteshini sokusekela ehle kusukela ekuqhumeni okungu-6 ngesonto lokuqala lamahemuhemu (akukho mnikazi, akukho nothi) kuya kumbuzo owodwa ocacisayo uma isixhumanisi se-Notion kanye nokuthunyelwe kwesiteshi sekukhona. Kuhlu lokuhlola lokufuduka (ukuthumela ngaphandle kuqediwe, iProjekthi ihlolwe, amadokhumenti avuselelwe, umnikazi uqanjwe igama, izimfihlo zisusiwe), ama-bots amathathu kwangu-3 akhiwe kabusha adlulile ngaphambi kokumemezela ngokumelene no-0 kwangu-3 ngaphansi komkhuba othi "sizobhekana nakho uma kuphula". Imikhawulo: iqembu elincane, indawo yomkhiqizo owodwa; amafayela ama-bots angenalo ulwazi ahamba ngokushesha; ama-GPT asebenza kanzima athatha isikhathi eside; isikhathi asifakiwe ekulindeni ukuvunyelwa kokuphatha kwama-Workspace Agents.

Ukuze ulinganise inguqulo yakho: faka uhlu lwazo zonke i-GPT ezenziwe ngokwezifiso ngamalebula abalulekile; ukuthumela isikhathi nokwakha kabusha i-bot ngayinye ebalulekile; landelela izinga lokudlula ohlwini lokuhlola kanye nezicelo zosizo ngemuva kokusika amasonto amabili; umbiko ubala ngama-denominator.

Yini engase ihambe kabi

• Silinde isimemezelo esiphelele: Yenza okuthile ngenkathi usakwazi ukuhlela nokulanda.
• Ukuthutha konke: Ukugcina amathoyizi; ukwakha kabusha amahhashi asebenzayo.
• Imiyalelo engenamafayela: Ingxenye yobuchopho. Thatha i-pantry, hhayi ikhadi leresiphi kuphela.
• Ukunqamula okuthule: Izixhumanisi ezabiwe ekungeneni ziba yi-attic enezipoki.
• Izimfihlo ku-wiki: Hlukanisa okhiye be-API uma unamathisela ukuthumela kwamanye ama-export kuma-shared drive.
• Uma sicabanga ukuthi amaphrojekthi afana: Hlola imiphumela, bese umemezela ukuthi usuqedile - hhayi ngenye indlela.
• Akukho mnikazi oqanjwe ngegama: "Ithimba" alikwazi ukuzungezisa i-URL efile.

Ukudla okuwusizo

Uma ama-GPT enziwe ngokwezifiso evala, okudingeka ukwazi ukuthi ayasebenza: isitokwe, imiyalelo yokuthumela ngaphandle namafayela olwazi, ukwakha kabusha abambalwa abathola imali yabo kumaphrojekthi e-ChatGPT (noma ama-plugin / Ama-Workspace Agents lapho amathuluzi nenqubo yeqembu kufaneleka kangcono), xwayisa noma ubani owabelane ngesixhumanisi esidala, bese weqa udoti. Umphefumulo we-bot uwumbhalo. Umbhalo uyahamba. Gcina izindlela zokupheka; shintsha ikhishi.

Imibuzo Evame Ukubuzwa

Kusho ukuthini ukuvalwa kwe-Custom GPT kubasebenzisi be-ChatGPT?

I-OpenAI iqeda ama-GPT Angokwezifiso kuzo zonke izinhlelo ze-ChatGPT. Ama-akhawunti e-Personal Free, Go, Plus, kanye ne-Pro awakwazi ukudala noma ukushicilela amasha, kuyilapho ama-GPT akhona engase asebenze noma ahlele isikhashana - aphathe lokho njengomusa wesikhashana, hhayi isithembiso. Ukusetha indawo yokusebenza kanye nebhizinisi kulandela indlela yokuqeda eshicilelwe eya kubalandeli abanjengama-plugin, ama-ChatGPT Projects, kanye nama-Workspace Agents. Yeka ukwakha ama-GPT amasha Angokwezifiso bese uqala ukusindisa lawo athola imali yawo.

Ngingakwazi yini ukudala noma ukushicilela ama-GPT amasha enziwe ngokwezifiso?

Kuma-akhawunti amaningi omuntu siqu, akukho - ukudalwa nokushicilela sekuvele kukhiyiwe, kufaka phakathi uhlu olusha lwe-GPT Store. Ama-GPT akhona enziwe ngokwezifiso onawo noma owabelwana ngawo angase avuleke, futhi kwezinye izinhlelo usengahlela imiyalelo noma amafayela olwazi ngaphambi kokuba lokho kufinyelela kunyamalale. Imisebenzi ebalulekile akufanele iphathe ama-GPT enziwe ngokwezifiso njengekhaya lesikhathi eside. Kopisha imiyalelo, iziqalo, namafayela ngenkathi usakwazi.

Ukushintshashintsha kwesimo sokuphumula kwezinhlelo zomuntu siqu kanye nezinhlelo zokusebenza kuhluke kanjani?

Abasebenzisi bomuntu siqu badinga kakhulu isipele somuntu siqu: ukuthumela imiyalelo namafayela, ukwakha kabusha kumaphrojekthi noma kwelinye ithuluzi, futhi baxwayise noma ubani owayenesixhumanisi esabiwe. Abaphathi bebhizinisi nendawo yokusebenza babhekene nama-bots abiwe ngaphakathi kokubhalisa, imibuzo nezimpendulo zenqubomgomo, amathrekhi okuthengisa, kanye nama-macro asekelayo - ngakho-ke isitoko ezingeni lenhlangano, nika abanikazi, bese ukhetha abazolandela ngaphambi kokuba lawo ma-bots ayeke ukusebenza. Uma uyisebenzisa kuphela, yenza isipele kuleli sonto; uma iqembu liyisebenzisa, liphathe njengokuncishiswa komkhiqizo omncane.

Yini okufanele ngiyifake ohlwini ngaphambi kokuba ama-GPT enziwe ngokwezifiso ayeke ukusebenza?

Kuwo wonke amagama nezinhloso ze-GPT, umnikazi, oncike kukho, ubude bemiyalelo, amafayela olwazi, izenzo noma amathuluzi, ukubaluleka, kanye nokuqagela okulandelayo. Hlunga ngokusetshenziswa okuphilayo - okubalulekile kwansuku zonke uma kuqhathaniswa nothuli lwedijithali - hhayi ngokukhumbula isikhathi esidlule. Abantu abaningi bathola ama-GPT amathathu enza umsebenzi ofanayo; hlanganisa ngenkathi ufuduka. Awukwazi ukuthutha lokho ongakwazi ukukusho.

Yiziphi izimpahla ze-GPT ezenziwe ngokwezifiso ezibaluleke kakhulu ukuzithumela kwamanye amazwe?

Imiyalelo yesistimu namafayela olwazi abalulekile - ubuchopho kanye ne-pantry. Kopisha futhi iziqalo zengxoxo, isenzo noma amanothi e-API ngolimi lwabantu ngaphandle kokunamathisela izimfihlo ku-Slack, izingxoxo ezimbalwa eziyisampula ezibonisa umphumela omuhle, uhlu lokwabelana ukuze abantu baxwayise, kanye nekhophi yohlu lwe-GPT Store uma ushicilele. Landa amafayela kufolda enegama elicacile nge-GPT ngayinye kanye nama-PDF emfihlakalo e-dual-backup avela ku-Drive uma izinguqulo zihlukile.

Ngingayakha kanjani kabusha i-GPT Yangokwezifiso kumaphrojekthi e-ChatGPT?

Dala iphrojekthi eqanjwe ngomsebenzi, unamathisele imiyalelo ethunyelwe kwamanye amazwe, ulayishe amafayela olwazi kuphela osawadingayo, bese uhlola izikhuthazo ezifanayo eceleni kwe-GPT endala. Mema abasebenzisi bangaphambilini bese ususa isixhumanisi esidala uma bekhululekile. Amaphrojekthi afanelana nomsebenzi oqhubekayo - ucwaningo, amapayipi okuqukethwe, izifinyezo zamakhasimende, imibuzo nezimpendulo zangaphakathi - ngaphezu kwezinhlelo zokusebenza zomphakathi ze-GPT Store. Qala ngama-bots akho amabili noma amathathu abalulekile, hhayi yonke i-zoo.

Kufanele ngisebenzise nini ama-plugin noma ama-Workspace Agents esikhundleni samaProjekthi?

Ama-bot anzima okufundisa anamafayela avame ukuhambisana namaProjekthi. Ithuluzi noma ukuzenzakalela kwe-API kuhambisana nama-plugin noma abalandeli besitayela sesenzo - qinisekisa kabusha ngokucophelela. Ithimba licubungula ngokulingana kobunikazi obucacile Ama-Ejenti Esikhala Sokusebenzela anomphathi onolwazi. Ama-bot okuthola umphakathi adinga ukucabanga kabusha ngoba inkathi yeSitolo se-GPT Esingokwezifiso iyaphela, futhi amathoyizi angenziwa kanye angagcinwa ngaphandle kokufuduka. Lindela ukungqubuzana okuthile; ama-clone omuntu nomuntu akunakwenzeka.

Yiziphi izingozi ezidalwa yiqembu elihlanganyelwe i-Custom GPTs ngesikhathi sokuvalwa?

Ama-GPT abiwe avame ukuhlala ku-Notion, amadokhumenti angena ngemvume, kanye ne-Slack engenamnikazi oqanjwe ngegama - bese ukuhlela kuyamiswa bese i-bot wonke umuntu ayisebenzisayo iba yi-attic enemizwa. Nika umnikazi womuntu, thumela okushintshayo nokuthi kufanele ushintshe nini, shintsha izixhumanisi eziqinile ngenkathi i-GPT endala isaphendula, bese ugcina ingobo yomlando yemiyalelo yokufunda kuphela. Kuma-bots abhekene namakhasimende, hlela umlayezo ozolile. Inqubo engenasipele yayiyithemba eline-interface yengxoxo.

Yimaphi amaphutha okufuduka okufanele ngiwagweme?

Ungalindi isimemezelo esiphelele - yenza ngesikhathi usakwazi ukuhlela nokulanda. Thutha okubalulekile nokusebenziseka kakhulu kuphela; gcina okusele kungobo yomlando. Ungakhohlwa amafayela olwazi, namathisela okhiye be-API kuma-wiki abiwe, dala kabusha ithoni phakathi kokufuduka, noma usike buthule. Amaphrojekthi awafani - hlola imiphumela ngaphambi kokumemezela ukuthi kwenziwe, bese ukopisha umbhalo wohlu lwesitolo manje uma kubalulekile. Yeqa amakomidi okuqondisa ukuze uthole i-bot esetshenziswa kabili ngenyanga.

Ngingayithutha kanjani i-GPT yokusekela okwabiwe ibe yiProjekthi?

Faka ohlwini i-bot, thumela imiyalelo ephelele kanye nefayela ngalinye lolwazi, londoloza iziqalisi kanye nezingxoxo zesampula, bese wakha kabusha kuphrojekthi enemithetho elungisiwe namafayela aqinisekisiwe. Hlola izimpendulo ezifanayo zokuhlola eceleni, buyekeza izixhumanisi ze-Notion noma zokufaka, bese wazisa isiteshi ngegama lomnikazi ngaphambi kokuba i-URL endala ife. Ukwamukelwa kusho ukuthumela ngaphandle okuphelele, ikhwalithi yesampula efanayo, amadokhumenti abuyekeziwe, futhi akukho zimfihlo ku-wiki yeqembu.

Izinkomba

1. I-OpenAI — Ukuqeda ama-GPT enziwe ngokwezifiso — help.openai.com