Why I-AI Agents kufanele usebenzise i-Mundane, ngakho-ke abantu akufanele

U-engineer no-innovator u-Matteo Zamparini, omunye ingxubevange elikhulu yokukhiqiza ehlabathini elidlulileyo ayikho isizukulwane se-talent noma isizukulwane se-ambiance, kuyinto isikhathi esithathwe ku-tasks ezivamile.

Ukusuka ukucubungula amaphepha amaphakheji njenge-data science intern kuya ku-co-founding a start-up eyenza izinzuzo ze-renewable energy kumadokhumenti amabhizinisi amabhizinisi amabhizinisi, umkhakha wakhe uye wahlala isihloko esiyingqayizivele: ukwakha ubuchwepheshe ukukhiqiza umsebenzi we-grunt ukuze umsebenzi wabantu angasetshenziselwa imisebenzi ezingaphezu kwe-strategic.

Njengomfundisi wokufunda we-PropRise, inkampani eyenziwe yi-Y Combinator, uye ukwakha ama-AI agents ezisebenza emhlabeni wonke ukujabulela inani elikhulu lwezimali ezahlukile zezimali kanye nokukhishwa kwezibuyekezo ezibonisa izinzuzo zokuthutha.

Konke okuhlobene ukubuyekeza kwelanga lapho i-AI ayikwazanga kuphela abantu, kodwa iveza kwabo ku-workforce ukuze zihlole izicelo eziningana.

I-Cost Real ye-Manual Data Analysis

Ngo-software-assisted world, ukunciphisa imishini ye-digital akuyona inkinga yokuqala. Ngaphandle kwalokho,UkuhlobisaI-California University of Irvine, i-California University of Irvine, i-California University of Irvine, i-California University of California, i-California University of California, i-California University of California, i-California University of California, i-California University of California, i-California University of California, i-California University of California, i-California University of California, i-California University of California, i-California University of California, i-California University of California, i-California University of California, i-California University of California, i-California University of California, i-California University of California, i-California University of California, i-California University of California, i-California, i-California,

Njengomfundisi we-internship, wahlala izinyanga ezisebenzisa izixhobo ezifana ne-maps ye-digital, ngokuvumela izindawo, izinhlelo zokusebenza ze-statistics kanye nezivakashi zebhizinisi zihlanganisa futhi ekupheleni kokufaka idatha ezisetshenziselwa ukuthatha izixazululo ze-investment ne-development. "Ngingathanda ukuchithwa: 'I-satellite imaging eyenziwe nge-Lidar kuyinto enhle,' "Ukuhlola. ' Kodwa 'Ukuhlola izinsuku ezinguqulo ku-maps kanye nezivakashi ukuhlola izindawo ezinhle zebhizinisi ngokuvamile kuyinto ukusetshenziswa kwe-intelligent yabantu?'"

Kuyinto enye ingcindezi elidlulile i-Spaceflare, i-startup eyenziwe ngama-UC Irvine abafundi abalandeli ngo-2021. I-model eyenziwe usebenzisa amacebiso zomphakathi kanye nama-satellite ukubonisa izindawo ezinhle zokufaka i-panel solar, okuvumela amabhizinisi be-energy kanye nama-property owners ukuhlola ngokushesha izinzuzo ezingenalutho zokukhiqiza amandla kanye nezimali.

I-process eyenza izinsuku ezingenalutho yokuhlola kwama-manual kungenziwa ngezinyathelo ezingu-minutes nge-Spaceflare.

Building AI Agents that Act and Reason—Within Reason

Namhlanje, i-Zamparini kuyinto nomkhiqizi emzimbeni we-AINgena ngemvumeLezi zebhizinisi zokusebenza njenge "i-AI investment analysts" ezingu-24 / 7 futhi zihlanganisa izindawo eziningi zebhizinisi ezingenalutho ezingenalutho ezingenalutho ezingenalutho ezingenalutho ezingenalutho ezingenalutho ezingenalutho ezingenalutho ezingenalutho ezingenalutho ezingenalutho ezingenalutho ezingenalutho ezingenalutho ezingenalutho ezingenalutho ezingenalutho ezingenalutho ezingenalutho ezingenalutho ezingenalutho ezingenalutho.

I-PropRise ayikho kuphela isixhobo se-machine learning esithathwayo yokuthengisa idatha; ama-agents zayo zokusebenza ngokuzimela, ukwakhiwa kwedatha etholakalayo futhi ngempumelelo ngokushesha ukubuyekeza nokunemba kwezilinganiso ezininzi. Zihlola izinzuzo ezithathelwe kumakhasimende wezilwanyana, kodwa angakwazi ukuguqulwa izixazululo ngokuvumelana kwabo ukuze zihlole izixazululo ezithakazelisayo.

"Sithuthukise izinhlelo zokusebenza ukuthi angakwazi ukuhlola, isebenza, futhi ngisho ukuhlola ukubuyekeza umuntu noma ukubuyekeza mid-task," wathi. "Ukuhlukaniswa okuzenzakalelayo kwezobuchwepheshe kanye ne-human oversight kuyadingeka ukukhiqiza i-AI workflows eqinile."

I-Human-in-the-loop approach iyinhlanganisela ebalulekile yokuqala lapho ama-AI agents angakwazi ukuthatha yonke umsebenzi wokubacindezela ngaphandle kokuphindaphinda ukuphazamiseka kwamahhala.

I-Reliability Over Novelty: Ukwakhiwa kwezinhlelo zokusebenza ze-Agentic ukuze kusetshenziswe kwihlabathi

Ngokungafani noZamparini, ukuthuthukiswa kwama-standards ezivela njenge-Model Context Protocol (MCP), i-AGNTCY, i-HumanLayer, noma i-language BAML kuyinto ingcindezi enhle kakhulu e-AI. Kulabo, zihlanganisa izakhiwo ezijwayelekile ze-generation elilandelayo ze-AI systems, futhi zihlanganisa imiklamo yayo yokwakhiwa.

Uma imiphumela ye-software ibhizinisi amamodeli yemali noma imibuzo yestrategic, imiphumela etholakalayo kufanele kuhlolwa. Ukuze idatha etholakalayo kuyinto ephelele, kufanele kube lula ukufinyelela.

Izinkqubo ezivela ezifana ne-HumanLayer, i-BAML, i-MCP, ne-AGNTCY zibonisa isisombululo esizayo ukuthi kungenziwa uhlobo le-ecosystem ye-agent.

I-HumanLayer, isibonelo, inikeza i-protocol enikezela ukubuyekezwa kwe-human-in-the-loop ku-agents e-AI. I-HumanLayer ikhipha i-actions e-high-stakes futhi i-routing ze-humans - nge-Slack, i-imeyili, i-SMS, nezinye izindlela-for approvals, i-feedback, noma i-escalation, bese i-feed-back ye-reaction ukuze ama-agents angafakwa ukusebenza ngokushesha.

I-MCP, i-standard ebonakalayo ebonakalisa indlela amamodeli ye-AI ukuxhumanisa ama-data inputs, okuvumela amamodeli amabili (i-LLMs) ukuxhumanisa ngqo amasethi amaningi kanye ne-API yokwenza isikhunta esikhulu se-information kanye ne-agent-to-agent collaboration.

I-AGNTCY kuyinto i-ecosystem epholile ye-agent-to-agent interoperability enikezela inkqubo ngokuvamile ngokuvumelana ne-shared identity layer kanye ne-interaction model ye-AI agents, okuvumela izixhobo ezivamile ezivamile ziye ziye zithunyelwe ukuxhumana kwebhizinisi ngokuphathelene futhi ngokuphathelene. I-AGNTCY inikeza "i-Internet of Agents", efana ne-shared data protocols eminyakeni angu-1990 eyenza ukuchithwa kwe-internet emhlabeni wonke.

Ngezinye, lezi zincwadi zibonisa izesekeli ze-AI ezinzima kuphela, kodwa zithunyelwe, zibonakalayo, futhi zihlanganisa.

Building a Template for the Future of Work

Izinzuzo ezahlukile nezilimi zihlanganisa ukuthuthukiswa kwe-AI okuyinto zingathola ukusebenza okuqhubekayo. Futhi zihlanganisa ngaphezulu kwe-commercial real estate; okungenani, zihlanganisa ama-AI ama-agents kanye nama-agent systems ukuguqulwa indlela abanamakhasimende kanye nama-experts abasebenza emhlabeni wonke.

“Nge-AI-assistance, singakwazi ngokuvamile ukuvikela isikhathi yabasebenzi,” Zamparini isho. Uma singakwazi ngokwenene ukuhumusha umsebenzi okuzenzakalelayo kanye nezixazululo ukuba zihlukile futhi zihlukile, izinhlelo zokusebenza ngaphezulu kunazo ukwandisa isikhathi esiyingqayizivele, zihlukile indawo izimboni ezintsha kanye nezinsizakalo.

Ukuze Zamparini, umlando we-AI ayikho umphathi we-digital assistant engcono, kuyinto uhlelo oluthumela abantu ukwakha izinto ezintsha ngokukhawuleza, ukuthuthukiswa kwezimboni ezivamile, futhi ngisho ukuguqulwa kwezimboni ezintsha. Ngaphandle kokuphumula kwezokukhiqiza, kuyinto ukuguqulwa kwezobuchwepheshe nezobuchwepheshe zemvelo.

I-AI ama-agents akuthunyelwe i-noise, ukunciphisa ukuxuba ngakho-ke ama-builders angakwazi ukujikeleza ngokukhawuleza, ukufikelela okungaphezulu, futhi ukhangela kakhulu.