Ngo-2026, umongameli we-United States we-Federal Ngemuva kokufaka izicelo zomthetho zihlanganisa izicelo zokusebenza zokusebenza ngokugcwele ziye zithunyelwe yi-artificial intelligence, inkulumo enhle ukuthi i-AI inokukwazi ukukhiqiza imiphumela emangalisayo okuyinto ngokugcwele ezahlukile emzimbeni. Ukuhlobisa iqembu le-advocates Ukuhlobisa iqembu le-advocates izinyanga ezedlule, abacwaningi be-cybersecurity babonisa ukuthi ingashintsha umphathi we-AI ukuze usebenza izicelo ezingenalutho ze-smart-home, okuvumela ukuthi izinhlelo zokusebenza okuzenzakalelayo zingatholakala lapho zihlanganisa ama-signals zangaphakathi. malicious calendar invitation Ukukhangisa i-kalendar Ngaphezu kwalokho, ama-threat intelligence analysts babe Ukubuyekezwa kwe-I.I. ye-agent-style ku-orchestra coordinated intrusion and extortion campaigns e-multi-organizations, okukhuthaza ukuthi izinhlelo zokusebenza ngaphandle kokuphumelela kokuphumelela kokuphumelela kwe-contextual. I-Documented Emerging Cyber Attacks I-Documented Emerging Cyber Attacks Lezi zincwadi zihlanganisa ama-anomali. Zihlanganisa isixazululo esikhulu se-enterprise: izinhlelo ze-AI ezibonisa ukujabulela okuhle kodwa zihlanganisa isikhokelo se-contextual esidingo ukusebenza ngokushesha futhi ngokuqhubekayo emkhakheni se-business. I-Cost Of Ignoring I-Context Imiphumela ye-ignorance ye-contextual accuracy iyatholakala ngokushesha. I-AI-generated code ebutholakalayo ingathuthukisa izinga lokuphumula noma izinhlelo. Ukusabela okuzenzakalelayo okuzenzakalelayo emkhakheni asebenzayo kungabangela ukuphazamiseka kwe-compliance. Kwamanye amazwe, izixhobo ze-AI zithunywe ngokushesha ngemuva kokuphumula kwe-enthusiasm ngokushesha ngenxa yokuba amabhizinisi asikwazi ukuvuselelwa kweziphumo zayo. Ukuze abasebenzi zobuchwepheshe, lokhu inikeza i-paradox emangalisayo: I-AI engcono kakhulu, kunzima kakhulu lapho isebenza ngaphandle kwe-Enterprise awuthanda. Ukungcweliswa kwe-context ikakhulukazi ikhiqiza i-downstream cascade ye-engineering friction. Amadivayisi we-AI abanjulile ukuxhumana kwezilinganiso ze-system noma umlando we-project angakwazi ukukhiqiza ikhodi ephakeme, okungenani okungenani, okuholela ama-pull requests, ukulungiselela ama-buggy, kanye namahhala amahora we-code review. Ngaphandle kokushicilela ukukhiqizwa, amaqela zithole isikhathi esikhulu ekubunjweni, ukuguqulwa, kanye nokuguqulwa kwe-AI outputs, okungenani elilodwa le-bottleneck ukuthi i-AI ye-contextual iyatholakala ukuguqulwa. Izinhlelo zokuqala zibonakalisa ukuthi amamodeli amabili (i-LLMs) angakwazi ukuphefumula imikhiqizo, ukuguqulwa ikhodi, nokuphendula imibuzo efanelekayo ngokushesha ngokushesha. Ukuvumela kulandelayo ngokushesha. Ku-2025, cishe Ukusebenza kwe-AI ku-generative ku-minimum ye-business function, okuvumela ukuthi i-AI iyahluka kusuka ku-experimentation kuya ku-operational infrastructure. Njengoba izimboni zithunyelwe izixhobo zokusebenza emkhakheni zokukhiqiza, isizukulwane esitsha esuka: Model Intelligence kuphela ayinempumelelo. Izinkampani zibonise ukuthi ingxubevange yokufakelwa kwe-AI ku-context epholile, esebenzayo ye-internal systems, workflows, kanye ne-institutional knowledge. Lokhu ukujabulela okuqhubekayo lithunyelwe isibambo se-innovation esekelwe ngaphandle kokwakha amamodeli amakhulu, kodwa ku-development ye-contextual understanding, nge-companies njenge- Ukuhlola izinhlelo zokusungulwa ukuhlanganisa idatha yobuchwepheshe, ikhodi, kanye nokuqinisekisa ku-intelligence layers esebenzayo. 67% Izinhlelo zokusebenza Nami 67% Izinhlelo zokusebenza Nami Ukukhishwa kwe-Enterprise AI Context Gap I-problem ikholele ekubeni ukuthi izinhlelo ezivela ekubeni ukuthatha i-LLM. Iningi kwezinkampani zihlanganisa amamodeli njenge-engine ye-knowledge eyahlukile, ngokuvumelana ukuthi ukuxhuma ama-documents e-internal ku-prompts ngeke ziye zihlanganisa kumasipala e-enterprise-ready. Ngokuvamile, amazwe amaningi zihlanganisa ngokwakha izibuyekezo zokushesha, ukuhlolwa nge-fine-tuning, noma ukwenziwa i-recieval-augmented generation (RAG) pipelines. Nakuba lezi zindlela zithuthukisa ngokushesha, zihlanganisa ngokuvamile njengesithombe esithathile kunazo zokusebenza. Nangona kunjalo, izimo zebhizinisi zebhizinisi zenzakalo. Zihlanganisa ama-ecosystems ze-codebases ezivela, izifundo ezincinane, izinhlelo zokuhweba, izitimela zokusebenza, kanye nezivumelwano zebhizinisi. I-LLMs zihlanganiswe ku-private data esikhulu sokuphendula lezi zebhizinisi ngokushesha ngaphandle kokuqinisekisa okuqhubekayo. Ukuhlolwa kwama-developer workflows kubonisa ukuthi izinhlelo ze-AI ezinemibuzo ye-enterprise-specific ngokuvamile zihlanganisa imiphumela ebonakalayo kodwa kuncike ukuvalwa okuhle, okuvimbela izinga lokukhiqiza okuyinto zibonakalayo. Yini ukwandisa inkinga Ukulungiswa kwe-AI ye-Enterprise kusebenza ngokushesha, kodwa isixazululo sokupholisa. Uma i-AI inikeza emhlabeni wonke Umphakeli we-Intelligent Artificial Intelligence (AI) umphakeli wabhalisela ukuthi akufanele ngokuphelele kwama-Intelligent Artificial Intelligence (AI) emaphaketheni, ngokuvamile usebenzisa isikhathi eside yokubuyisa noma ukuguqulwa kwe-Intelligent Artificial Intelligence. 42% of committed code 42% ye-committed code Futhi izinhlelo eziphambili ze-AI zihlanganisa: a Ngaphansi kwe-45% yezimpendulo ze-AI assistant zihlanganisa i-inaccuracies ezinhle, kusuka kufakazi ezingenalutho kuya kuma-attributions ezingenalutho. isifinyezo isifinyezo I-Data ye-Enterprise e-Fragmented eyenziwe nge-CRMs, izinhlelo zokuphefumula, izifundo ze-knowledge, kanye ne-legacy logs ikhiqiza ukuthi abacwaningi abizwa ngokuthi , okuvumela i-AI ukuba awukwazi ukuguqulwa kwezilinganiso noma izilinganiso. Amandla Amnesia Amandla Amnesia Izinzuzo zokhuseleko zihlanganisa ingozi. Abacwaningi akhawunti ukuthi ikhodi eyenziwe yi-AI angakwazi Ukwenza i-"Illusion of Correctness" okuvimbela imiphumela yokwakhiwa kwezakhiwo lapho abavela abavela ku-validate imiphumela ngokugcwele. Ukubhalisa Izinzuzo Ukubhalisa Izinzuzo Izinzuzo zithunyelwe ukuthi izixazululo ze-context, ukuphazamiseka kwe-trust, kanye ne-verification load zithunyelwa kwezimpendulo ezikhuthazayo, zihlanganisa amabhizinisi ukuhlangabezana imiphumela ye-AI ngaphandle kokushesha i-workflows. Indlela Izinkampani Zokusebenza Ukuqhathanisa It Izindlela ezininzi ziye ziye ziye ziye ziye ziye zithunywe njengezinkimbinkimbi zokusebenza ukunciphisa isixazululo. Ukuhlobisa amamodeli ku-internal data kubuyekeza ukubuyekezwa kwe- vocabulary kodwa ibhimbele ukugcina imizamo embalwa ngokushesha. Ukuhlobisa okusheshayo ibhimbela ukuxhumana neamodeli kodwa ayikwazi ukuguqulwa imibuzo embalwa phakathi kwama-system kanye ne-workflows. I-RAG ipayipi ezivamile ibhimbele ukufinyelela kwebhizinisi kodwa ikakhulukazi asikwazi ukujabulisa ukuxhaswa, ukuxhaswa, kanye nokuqiniseka kweziqhingi. Ngaphezu kwalokho, izinhlelo zokusebenza nezinhlelo zokusebenza zokusebenza nezinhlelo zokusebenza ezijwayelekile zihlanganisa ulwazi se-Enterprise njenge-diagramm esebenzayo yokuxhumana phakathi kwe-code, i-documentation, i-tickets, ne-signals zokusebenza. Izindawo ezifana neNaboo zihlola le kategorie ngokusekelwe izindwangu zokusebenza zokusebenza ezingenalutho ezivumelanisa izinhlelo zokusebenza ze-AI ukuxhumana nezinhlangano kunoma kuphela ukulanda izidakamizwa. Ngokuvimbela izinhlelo zokusebenza zebhizinisi, izixhobo zokuxhumana, kanye nezinhlangano zokuthuthukiswa ku izakhiwo ze-semantic ezivumelanayo, izinhlelo zihlanganisa i-hallucinations kanye nokuvumela ama-AI agents ukwenza imisebenzi zokucwaninga asebenzayo. Ukulungiselela kanjani imiphumela emhlabeni wonke Ukuze izidakamizwa izidakamizwa, imiphumela ikakhulukazi ibonisa isikhathi sokunciphisa umjikelezo. Iziqu ze-AI izesekeli ziye zitholele imibuzo mayelana nezinhlangano zangaphakathi, imibuzo yokwakhiwa, noma i-ticket history nge-level ye-accuracy eyenza izinzuzo zangaphakathi zenzuzo. Izixhobo zihlanganisa izinzuzo ezahlukile: I-AI ebonakalayo kuphela i-syntax, kodwa ukuhlola ukusebenza kwe-system. Uma i-context iyahlekile ngokuqhubekayo, i-AI ingatholela izibuyekezo ze-code ezihambelana ne-architecture eyenziwe, ama-test frameworks, ne-compliance requirements, ngokuvamile ukunciphisa i-rework. Kwi-executive level, I-AI ye-contextual inikeza imiphumela eningi: ROI enesilinganiselwe. Umongameli inikeza ukubukeka ku-development progress, i-tech debt trends, ne-operational risks ngokusebenzisa izinhlelo zokusebenza ze-AI ezibonisa ama-signals ezizodwa zebhizinisi kunokuba ama-benchmarks zezimboni. Ngaphandle kwe-metrics, izinhlelo zokusebenza zihlanganisa ama-documents ezithile, i-code, ne-dependances ezinxulumene ne-task, ukunceda abasebenzi ukholelwa ukuthi akuyona kuphela, kodwa kanjani, futhi lapho ukunakekelwa noma ukuxhumana kuyadingeka. I-Opportunity Ngaphezulu kwe-Problem Uma amabhizinisi bakwazi ukuguqulwa kwe-contextual challenge, ukucubungula okwenziwe ngempumelelo kunezinguqulo zokukhiqiza. Amakhasimende we-AI angakwazi ukugcina izinhlelo ezithakazelisayo ngokuhambisana, ukucubungula i-software modernization, nokuvumela amabhizinisi amabhizinisi ukucindezeleka ikakhulukazi ku-innovation kunoma ku-system navigation. Izinhlelo zingathola ulwazi lwezinhlelo zokusebenza okuyinto zihlanganisa izixhobo ezincinane kanye nezidakamizwa ze-tribal. Abasebenzi abatsha zingatholakala ngokushesha. Ukusebenzisana kwe-cross-functional kungase kuthathwa nge-less translation layers phakathi kokuqinisekisa kwebhizinisi nokuhambisana kwezobuchwepheshe. I-future ye-AI yebhizinisi ingaphakathi ingatholakala ngokuvamile ku-building amamodeli amakhulu futhi ku-building infrastructures emangalisayo emkhakheni. Izinkampani ezivamile nge-contextual indexing platforms njenge-Naboo zihlanganisa ukuthi ukuxhumanisa ulwazi lwezinkampani, ikhodi, kanye ne-intention kungenziwa kwe-AI kusuka ku-assistant emangalisayo ku-operational collaborator enhle. Njengoba amabhizinisi asebenza ekuphuculeni ukusetshenziswa kwe-AI, izinhlangano ezihlabathi ziye zibonisa ukuthi ama-agents zabo zisebenza ngaphezulu kuka-generate i-response; futhi zihlola izilinganiso lapho izimpendulo zihlanganisa ukusebenza. 
 
 Lesi sihloko lithunyelwe njenge-release nguJon Stojan ngaphansi kwe-HackerNoon's Business Blogging Program. Lesi sihloko lithunyelwe njenge-release nguJon Stojan ngaphansi kwe-HackerNoon's Business Blogging Program.

The is an opinion piece based on the author’s POV and does not necessarily reflect the views of HackerNoon.

Lo msindo ukhiqizwa ngolimi lokuqala lwendaba!

I-Agents Without Context Yenza Imibuzo

About Author

IMIBONO

HANG TAGS

LESI SIHLOKO SETHULWE NGAPHAKATHI

Related Stories

Zulu Network: Moving The Bitcoin Economy Forward With a Two-Tiered Bitcoin Layer 2 Architecture

The Billionaires F*cked Up

Zulu Network: Moving The Bitcoin Economy Forward With a Two-Tiered Bitcoin Layer 2 Architecture

The Billionaires F*cked Up

Light-Mode

Classic

Newspaper

Minty

Dark-Mode

Neon Noir

Minty

HN StartUps