Ngiyaqala ngathi: I-AI kanye nezinhlangano zayo ezintsha ku-integration ziquka ingxaki, akuyona kumakhasimende, kodwa kumadivayisi abacindezelayo futhi abacindezelayo. Ngenxa yalokho? Ngenxa yalokho kuqala ukuxuba kuzo zonke izindawo ezithile ze-automation space futhi ngokushesha kodwa ngokushesha ziye ziye ziye ziye ziye ziye zihlabathi. I-AI technology entsha, njenge-MCP ne-ACP, iyatholakala kakhulu futhi, ngezindlela eziningi, kungenziwa ngokuvumelana nezinsizakalo zomuntu, uma siphinde sibe isikhathi sokusebenza. Yini entsha ku-Agentic AI? Okwe, ngingathanda ukuthi uma usofundisa lokhu, unayo ucwaningo olufanelekayo ye-agent AI, futhi uma akuyona, lapha isithombe se-agent. I-agent AI inguqulo oluphumelela ye-AI ukuhlanganiswa lapho singakwazi ukuhlola izixhobo kanye nezinsizakalo zokusebenza ukuthi i-LLMs zithumela ukusebenza kanye nokusebenza izinsizakalo. Izinsizakalo zingase zihlanganisa izakhiwo ze-API zangaphandle noma zangaphakathi, ukulayisha amafayela, ukufinyelela kwe-command layer esizodwa, noma ngisho ukufinyelela kwe-code. Izinto ezinhle kakhulu. Ngaphezu kwalokho, isicelo esisodwa se-inthanethi se-inthanethi se-inthanethi se-inthanethi se-inthanethi se-inthanethi se-inthanethi se-inthanethi se-inthanethi se-inthanethi se-inthanethi se-inthanethi se-inthanethi se-inthanethi se-inthanethi se-inthanethi se-inthanethi se-inthanethi se-inthanethi se-inthanethi se-inthanethi. Yini MCP? I-MCP, noma i-Model Context Protocol, kuyinto i-protocol ebonakalayo futhi ebonakalayo kanye ne-open-source definition for implementing or defining agentic tools and resources. Why is this a necessary thing? Ngithombe i-standard ebonakalayo yokusebenza kanye nesakhiwo ebonakalayo yokusebenza kanye nokuthuthaza izindlela kanye nezinsizakalo, okuyinto ngithanda T&R, kumakhasimende amaningi. Ngoku, ku-MCP, izinhlayiya ezimbili eziyinhloko zihlanganisa futhi zithole T&R kumakhasimende. Kukhona i-MCP server ne-MCP clients. I-MCP servers zihlanganisa imishini ye-host noma i-hub eziyinhloko lapho i-T&R ebonakalayo iyatholakala kumakhasimende ngezindlela docs noma i I-MCP client, Ngaphandle kwalokho, i-Agent futhi ingatholakala ku-MCP server futhi ukufinyelela izixhobo. I-server ikakhulukazi i-execution kanye ne-data procurement platform ye-agent futhi ayikho i-LLM ukufinyelela kuze kube usebenzisa into efana ne-Context (bheka i-docs). Ngena ngemvume Ukubuyekezwa Okunye, i-TLDR ye-flow ephakeme iyona i-agent ibiza i-server izifundo ze-tool, i-server ibonise izifundo ze-tool, futhi i-agent ibonise ukuthi umkhakha we-tool uye. Manje, lapho umsebenzisi akufundisa izinto ezihlobene nokusebenza kwe-tool, i-agent ibonise i-tool (ukufundisa i-server ukuba isebenze) futhi isetshenziselwa impendulo lokugqibela ukuze wabelane umsebenzisi ukuthi imiphumela ekupheleni. Njengoba ngithi, izinto ezangalisayo. Yini i-ACP ? Thola ku-MCP, futhi uye wahlala indlela yakho yokufunda i-docs futhi wahlala ukufundisa lokhu. Ngiyazi. Ngiyazi le-execution ephakeme, uma sincoma indlela yokwenza futhi sicela ama-agents eziningana, ngamunye nge-personas, izixhobo ze-MCP, ama-flows, kanye nama-capabilities ezahlukile, angakwazi ukuxhumanisa ukwenza umsebenzi, nje ukubonisa inkulumo yokusebenza okuvumela. Kuyinto i-ACP. I-ACP, noma i-Agent Communication Protocol, kuyinto i-open-source kanye ne-community-managed standard yokuvumela ukuxhumana kwe-agents. Okunye, kanjani okufanayo? It is a layer of implementation above the agents that is compatible with almost all agent definition SDKs out there. Yini kungcono ukuthi it defines a flow of execution, noma fixed noma flow yama-hierarchical, ukuxhuma ama-agents. Ngicabanga ukufinyelela ngokugcwele mayelana ne-fixed flow ( ), kodwa umphumela we-hierarchical kuyinto ingxenye eyahlukile ngokugcwele. Ungathanda nje njenge-map ye-agent e-multiple eyenziwe ngu-agent eyodwa ebizwa ngokuthi i-router agent. Ungathanda njenge-API, kodwa nge-brain, futhi enhle. Okunye docs Ngakho-ke inqubo kuyinto efanayo: umphathi we-router uzama zonke ama-agents ezihambelana ngezithombe zabo, futhi bonke ama-agents abasebenzi abahlanganisa imininingwane zabo ezifana ne-execution info, izivakashi ezihambelana, kanye nezixhobo ezifuywayo. Ngokuya, kulingana ne-user input, umphathi we-router uzama umphathi we-best for the job futhi uzame ukwenza umsebenzi. Futhi uma kukhona ama-agents eziningi ezihambelana ukwenza umsebenzi, uzama ukwenza umphathi lapho ngamunye umphathi iyatholakala umsebenzi, futhi umphathi we-router uzama ukuhlangabezana ulwazi, ukubuyekeza imiphumela kanye nemiphumela, futhi uzama umphumela kubasebenzisi. Ngemuva kokufuna lokhu, okungenani ku-surface level, Ngingathanda kakhulu kumadivayisi (ngibhalisa ingozi elilodwa kulesi). Singatholakala kwebhizinisi nge-cybersecurity. Ungayifaka kanjani indawo yokusebenza kwe-cybersec futhi isebenza ngokuphelele? Ngamanje, uma amazwe akufanele izicelo ezincinane ezivamile ziye zihlanganisa, zihlanganisa abasebenzi abesifazane abalandeli abalandeli abalandeli abalandeli abalandeli abalandeli abalandeli abalandeli abalandeli abalandeli abalandeli abalandeli abalandeli abalandeli abalandeli abalandeli abalandeli abalandeli abalandeli abalandeli abalandeli abalandeli abalandeli abalandeli abalandeli abalandeli abalandeli abalandeli abalandeli abalandeli abalandeli abalandeli abalandeli abalandeli abalandeli. Izinombolo zithembisa: Nge Ngaphezu kwalokho, i-96% ye-IT post-graduate inikeza iminyaka eminyakeni eminyakeni eminyakeni eminyakeni eminyakeni eminyakeni eminyakeni eminyakeni eminyakeni eminyakeni eminyakeni eminyakeni eminyakeni eminyakeni eminyakeni eminingi. I-55% kwezinqubo zokusebenza zokusebenza zokusebenza zokusebenza zokusebenza zokusebenza zokusebenza zokusebenza zokusebenza zokusebenza zokusebenza zokusebenza zokusebenza zokusebenza zokusebenza zokusebenza. I-94% yama-IT yokufinyelela izinga lokufinyelela okungenani eminyakeni Umphumela umphumela umphumela umphumela: , futhi i-41% yabasebenzi zihlanganisa ukunciphisa abasebenzi ngaphakathi kweminyaka engu-5 ngenxa ye-AI. Kwi-cyber security, Ngaphezu kwama-90% yama-Tier-1 i-automating. Ngaphandle kwe-Tier-1 , amabhizinisi ukhethe AI phezu amabhizinisi amasha. I-Big Tech ukunciphisa ukuqeqeshwa kwezidakamizwa ezintsha nge-25% ku-2024 I-AI uzokwandisa ukusebenza kwe-SOC ng-40% ku-2026 I-cybersecurity isixazululo se-4 million Yini ngibuyekeza ukuba abe ngaphambi kwe-curve? Qhagamshelana ne-AI ne-AI automations, uma ungenza 'em, qhagamshelana 'em. Ungafundisa ngezinto zayo futhi ukwandisa ulwazi lwakho kubo ngoba, njengoba kulokhu, izindleko zangaphakathi, futhi ngakho-ke i-input bar iyanda futhi i-automation ikhiqiza. Ngakho-ke, uma ungenza ngokushesha, ungakwazi ukuxhuma ibhizinisi futhi usebenzisa ikhono lakho lwezobuchwepheshe ukufaka i-automation kanye nokwenza isikhokelo esizayo. Ngithanda lokhu kuyimfuneko, kodwa asikwazi ukunceda. Ngaphezu kwalokho, izikhokelo ezinkulu ezinikezele ngempumelelo zihlanganisa, kodwa ama-blue ne-red teammers zihlanganisa futhi ngaphakathi kweminyaka eminingi, uzothola futhi ku-risk. Ngaphandle kokuphumelela kwe-AI, ucwaninga ukuba abe nomdlavuza we-"Human-in-the-Loop". Izinkampani zithanda abantu abanolwazi ukuhlola iziphumo ze-AI, ukuhlangabezana nezimo ezingenalutho ezingenalutho ezingenalutho ezingenalutho, futhi zihlale izinto lapho isofthiwe se-AI ibonise izinketho ezinzima. Lokhu kuyinto umugqa omtsha olulodwa lapho ungenza isisindo sakho ngokuvumelana ne-factor esiyingqayizivele yabantu enikezela i-AI ekuphumeni. Enye indlela kuyinto ukucindezeleka ukucindezeleka kwama-problem-solving kanye nezindawo lapho I-AI ibheka ebuthakathaka. Ukusabela okuqhubekayo, ukucindezeleka okuqhubekayo kwezimpendulo, ukucindezeleka kwezobuchwepheshe zokhuseleko, kanye nokuhlolwa kwestrategical zihlanganisa umntu, ukucindezeleka, kanye nokucindezeleka ngaphandle kwezimpendulo ezivamile. Imikhiqizo ye-AI ingasebenza ezemidlalo, kodwa ayikwazi ukucindezeleka kwezimpendulo ze-humane ngesikhathi sokuphendula ama-vektor ezintsha ze-attack noma ukuthuthukiswa kwezinkqubo zokhuseleko ezidingekayo ekubambisana nezidingo zebhizinisi. Ungasebenzisa kuphela izixhobo ze-AI, ukucindezel I-professional enhle kakhulu iyathuthukise iziqinisekiso zokusebenza, zihlanganisa ubuchwepheshe zokusebenza nge-business expertise, i-communication, ne-leadership. Yenza umbhali phakathi kwezindiza ze-tech kanye nezinhlangano zebhizinisi, into ye-AI ayikwazi ukwenza ngenxa yokuxhumana kwezobuchwepheshe zomuntu, i-policy, kanye nezinkinga ezivamile. Ekugcineni, ucwanjiswa kokusebenza ngokuvamile. Yenza i-AI-driven security products njenge-business esihlalweni. I-barrier yokufunda izixhobo ze-AI ayikwazanga engaphakathi, futhi ukuhlolwa kwama-value ngokusebenzisa ukwakhiwa kunokuba yindlela enhle yokuhlukanisa ngezikhathi eziningana nezinsimbi. Ukuphendula Ukuphakama Ngicabanga, kodwa lokhu kungcono. Thina kufuneka ukhusele izindleko zethu kanye nesikhathi esizayo se-cybersecurity. Umphumela wethu lokugqibela njengama-profesionists we-security ayikho kuphela ukhusela ulwazi, kodwa nathi. Kuyinto konke ukuthi ngithi,

The writer was physically present in relevant location(s) to this story. The location is also a prevalent aspect of this story be it news or otherwise.

Personal Website

LinkedIn

HuggingFace

Github

Cybersecurity Engineer

Lo msindo ukhiqizwa ngolimi lokuqala lwendaba!

I-Junior I-Cybersecurity Roles Is Avoid - I-Blame Agent AI

About Author

IMIBONO

HANG TAGS

LESI SIHLOKO SETHULWE NGAPHAKATHI

Related Stories

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

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

Light-Mode

Classic

Newspaper

Minty

Dark-Mode

Neon Noir

Minty

HN StartUps