- Indlela yekofu ivumela abaphuhlisi ukuba banikeze imisebenzi ephindaphindiweyo kwiiarhente ze-AI ezizimeleyo.
- Ukudityaniswa komthonyama kunye neKhowudi yeVS, iJetBrains, kunye nezixhobo ezifana neGitHub, iJira, kunye neSentry.
- Ukusebenza okugqwesileyo kwiibenchmarks eziphambili ezifana ne-SWE-Bench kunye ne-SWE-Lancer enkosi kwi-Repo Grokking teknoloji.
- Imodeli yexabiso eguquguqukayo kunye neenketho zasimahla kunye neemodeli zokuhlawula ezihlawulweyo kumashishini.
I-automation kuphuhliso lwesoftware ithathe umtsi obalulekileyo phambili ngokufika kwe Iiarhente zobuntlola ezenziweyo zeZencoder. Le nkampani esekelwe eSan Francisco iye yazisa isisombululo esingaphuculi imveliso yomdwelisi kuphela, kodwa yazisa into entsha: 'Indlela yeKofu', que ivumela abaphuhlisi ukuba banikeze ukuveliswa kwekhowudi kunye novavanyo lweyunithi kwi-AI ngelixa bethatha ikhefu.
Iiarhente ezintsha zeZencoder ziyilelwe ukudibanisa ngqo kwindawo ezidumileyo zophuhliso ezifana neKhowudi yeVisual Studio kunye neJetBrains., ngaloo ndlela ukuphepha ukutshintshwa kwezixhobo ezihlala zisebenza njengesithintelo ekwamkelweni kwezinye izisombululo ezifanayo. Oku kubonisa i-advanteji ephawulekayo ngaphezu kwabakhuphisana nabo njenge-Cursor, efuna ukusetyenziswa kwe-IDE ethile ukuze isebenze.
Iiarhente ze-AI eziqondayo ukuhamba kwakho komsebenzi
Endaweni yokunyanzela abaphuhlisi ukuba baphume kwindlela yabo yangoku, I-Zencoder iqhelana nemeko yokusebenza esekiweyo. Ngu ngaphezu kwama-20 okudityaniswa komthonyama Ngezixhobo eziphambili ezifana ne-GitHub, i-GitLab, i-Jira, okanye i-Sentry, ii-arhente zinokubandakanyeka kuwo wonke umjikelezo wophuhliso, ukusuka ekucwangcisweni ukuya kuqinisekiso lomgangatho.
Olunye udibaniso oluphawulekayo liqhosha elikhoyo kwiJira ebizwa “Sombulula ngeZencoder”, ekuvumela ukuba usombulule imiba ngokuzenzekelayo kwisixhobo solawulo ngokwaso ngaphandle kokutshintsha iifestile okanye iimeko.
Ubukrelekrele bomxholo yenye intsika yobu buchwepheshe.. Inkqubo isebenzisa into ebizwa ngokuba yiRepo Grokking™, evumela iiarhente ukuba zihlalutye nzulu uvimba weprojekthi yonke, ukuqonda ubume bayo, ukuxhomekeka, kunye nesimbo sekhowudi. Le ndlela ithintela imodeli ekuveliseni izisombululo ezinqanyuliweyo okanye ezingahambelaniyo nenkqubo yendalo yeprojekthi..
Imo yekofu: Thumela uvavanyo lokubhala ngelixa urhabula ikomityi yekofu.
Into ethe yatsala umdla woluntu, ngaphandle kwamathandabuzo, lukhetho Indlela yekofu. Lo msebenzi ivumela ababhekisi phambili ukuba benze i-arhente ezimeleyo ngonqakrazo olunye ukwenza ikhowudi okanye uvavanyo lweyunithi ngelixa umntu ethatha ikhefu.
Ngamazwi omseki uAndrew Filev, “Kufana nokuba nomntu osebenza naye onokuqhubeka ebhala iikhowudi ngelixa uphefumla.”. Umbono kukukhulula abaphuhlisi kwimisebenzi ephindaphindwayo efana novavanyo lokubhala, oluthi ngokubanzi alukhuthazi umdla phakathi kweqela kodwa luyimfuneko ekuqinisekiseni uzinzo lwesoftware.
I-arhente ayivelisi kuphela ikhowudi, kodwa iqinisekisa, ivavanya kwaye iyicokise ngokuzimeleyo.. Ngenxa yesakhono inkampani eye yabiza indlela yokuziphilisa, inkqubo ijonga ukuba isiphakamiso siyasebenza na phambi kokuba sisibuyisele kumsebenzisi.
Iziphumo eziphezulu kwiibenchmarks eziphambili
Ukusebenza kwezi arhente akupheleliselwanga kulwazi lwe-anecdotal.. I-Zencoder ipapashe iziphumo zayo kwiibhentshi ezininzi kwaye amanani abetha: iiarhente ziye zakwazi ukusombulula. I-63% yeengxaki eziphakanyiswe kumgangatho oQinisekisiweyo we-SWE-Bench, kunye nendlela esebenzayo yendlela enye-oko kukuthi, ngaphandle kokuvelisa iinzame ezininzi ngokuhambelanayo njengezinye, abakhuphisana abajolise ngakumbi kuphando benza.
Okuphawuleka ngakumbi kukuba kolona phawu lwamva nje lwe-SWE-Bench Multimodal, izinga lempumelelo lifikelele kuma-30%, kabini awona manqaku angcono arekhodiweyo ukuza kuthi ga ngoku. Kwi-Diamond ye-SWE-Lancer IC, ithathwa njengocelomngeni ngokukodwa, ii-arhente zibonisa ukusebenza okungaphezulu kwe-30%, ukuphucula ngamanqaku epesenti ye-20 kwisiphumo sangaphambili esingcono, kubandakanywa ne-OpenAI.
Yenzelwe abaphuhlisi bokwenyani
Uyilo lweearhente lujolise kwiimfuno zomphuhlisi oyingcali.. Isixhobo sixhasa ngaphezu kweelwimi ezingama-70 zokucwangcisa kwaye sigxile kwiindawo eziqhelekileyo zeshishini elinjengeJava kunye neC #.
NgokukaZineng Yuan, injineli eyintloko ePaytm, "Kufana nokuba nentanga ekulindeleyo into oyifunayo kwaye ikusindise kwimisebenzi ephindaphindiweyo. Ngethuba lokunikezelwa okubalulekileyo, wandinceda ukuba ndinqumle umsebenzi wam ngesiqingatha ngokucebisa isicwangciso sokubuyisela imodyuli yelifa.".
Inkampani nayo ithathele ingqalelo ukhuseleko. Yonke ikhowudi eveliswayo ihamba ngokuphononongwa okuzenzekelayo okubona ubuthathaka kunye nokuthobela imigangatho yokuthunyelwa kwemveliso.. Oku kuqiniswa ngokuhlolwa kokhuseleko oludibeneyo kunye neenkqubo zokuvavanya ezizenzekelayo ezinokuthi zisebenze zombini kwimibhobho ekhoyo kunye nabameli ngokwabo.
Ikamva langoku kunye namaxabiso
Imephu yendlela yeZencoder iqikelela inkqubela engakumbi unyaka wonke. Inkampani ngokwayo iqinisekisa ukuba ngaphambi kokuphela kuka-2025 siza kubona isizukulwana esitsha sabancedisi beekhowudi abanobuchule ngakumbi.
Ukwenza lula ukufikelela kwisixhobo sayo, iZencoder ikhethe imodeli enenqanaba elibandakanya:
- Inguqulelo yasimahla ngemisebenzi esisiseko.
- Isicwangciso seshishini i-$ 19 ngomsebenzisi ngamnye ngenyanga ngenkxaso yovavanyo oluphambili kunye nokuveliswa kwekhowudi ehlakaniphile.
- Isicwangciso seshishini i-$ 39 ngomsebenzisi ngenyanga, ebandakanya inkxaso ephambili kunye neempawu zokuthotyelwa eziphambili.
Ngolu khetho, inkampani ifuna ukuziqhelanisa namaqela amancinci kunye neenkampani ezinkulu. abafuna ukwandisa uphuhliso lwabo ngaphandle kokuncama iimeko zabo zangoku kunye nezixhobo.
Ukuphehlelelwa kweZencoder kubonisa inguquko ebonakalayo yophuhliso oluncediswa yi-AI. Kude nokuba nethemba eligqithisileyo okanye indlela yokungafuni, isisombululo sibonakala sifumana ulungelelwaniso lwepragmatic: ukuxhobisa abantu apho banempembelelo enkulu kunye nokubakhulula kumthwalo ophindaphindiweyo.. Konke oku ngaphandle kokushiya iindawo eziqhelekileyo zokusebenza okanye ukubeka esichengeni ukhuseleko lwemveliso yokugqibela.
Ndingumntu othanda itekhnoloji ojike umdla wakhe we "geek" waba ngumsebenzi. Ndichithe ngaphezulu kweminyaka eli-10 yobomi bam ndisebenzisa itekhnoloji yokusika kwaye ndikhenkceza ngazo zonke iintlobo zeenkqubo ngenxa yokufuna ukwazi okumsulwa. Ngoku ndiqeqeshelwe ubugcisa bekhompyutha nakwimidlalo yevidiyo. Oku kungenxa yokuba ngaphezu kweminyaka emi-5 ndibhalela iiwebhusayithi ezahlukeneyo kwitekhnoloji kunye nemidlalo yevidiyo, ndisenza amanqaku afuna ukukunika ulwazi oludingayo ngolwimi oluqondakalayo kuye wonke umntu.
Ukuba unayo nayiphi na imibuzo, ulwazi lwam lusuka kuyo yonke into enxulumene nenkqubo yokusebenza yeWindows kunye ne-Android yeefowuni eziphathwayo. Kwaye ukuzinikela kwam kukuwe, ndihlala ndikulungele ukuchitha imizuzu embalwa kwaye ndikuncede usombulule nayiphi na imibuzo onokuba nayo kweli lizwe le-intanethi.