Ukuhlanganisa i-Agentic AI ku-Software Development Workflows: Umhlahlandlela Osebenzayo

Ukuhlanganisa i-Agentic AI ku-Software Development Workflows: Umhlahlandlela Osebenzayo

Isingeniso

Ukwakheka kwesoftware kuthuthuka ngokushesha, ngokwanda kokwamukelwa kobuchwepheshe be-Artificial Intelligence (AI) kanye Nokufunda Ngomshini (ML). Enye yezindawo ezijabulisa kakhulu zocwaningo lwe-AI i-Agentic AI, egxile ekudaleni ama-agent ahlakaniphile angaxhumana nabantu kanye namanye ama-ejenti ukuze afinyelele imigomo eyinkimbinkimbi. Kulesi sihloko, sizohlola ukuthi abaklami be-software nabaphathi bephrojekthi bangasebenzisa kanjani amathuluzi e-AI e-agent afana ne-LangChain ne-DSPy ukuze kuthuthukiswe ukusebenzisana, izinqubo ezizenzakalelayo, kanye nokuthuthukisa ukulethwa kwesofthiwe ezindaweni zesimanje zokuthuthukiswa.

Iyini i-Agent AI?

I-Agentic AI ibhekisela ohlotsheni lwe-AI oludizayinelwe ukuthi lusebenzele abantu noma amanye ama-ejenti ukuze kuzuzwe izinjongo ezithile. Lawa ma-ejenti angasebenzisana nabantu, amanye ama-ejenti, kanye nemvelo ukuze aqoqe ulwazi, enze izinqumo, futhi athathe izinyathelo. I-Agent AI yehlukile ku-AI yendabuko ngoba yakhelwe ukuthi isebenze futhi izimele, kunokuba imane isebenze.

Izinzuzo ze-Agentic AI Ekuthuthukisweni Kwesofthiwe

Ngakho-ke, kungani abaklami be-software nabaphathi bephrojekthi kufanele bakhathalele i-AI ye-agent? Nazi ezinye zezinzuzo zokuhlanganisa i-AI ye-agent ekugelezeni komsebenzi wokuthuthukisa isofthiwe:

Ukusebenzelana Okuthuthukisiwe: I-Agentic AI ingasiza amaqembu ukuthi asebenzisane ngempumelelo ngokwenza imisebenzi ezenzakalelayo, inikeze impendulo yesikhathi sangempela, kanye nokwenza lula ukuxhumana.
Izinqubo Ezizenzakalelayo: I-Agent AI ingakwazi ukuzenzela imisebenzi ephindaphindwayo nedla isikhathi, ikhulule onjiniyela ukuze bagxile emisebenzini yezinga eliphezulu.
Ukwenza Izinqumo Okuthuthukisiwe: I-Agent AI inganikeza imininingwane eqhutshwa idatha nezincomo ukuze isize amaqembu enze izinqumo ezingcono.
Ukusebenza Kakuhle: I-Agentic AI ingasiza amaqembu asebenze kahle kakhulu ngokuhlonza izingqinamba, ukuthuthukisa ukugeleza komsebenzi, kanye nezinqubo zokulula.

I-LangChain: Uhlaka oludumile lwe-Agentic AI

I-LangChain wuhlaka lwe-AI oludumile oluvumela abathuthukisi ukuthi bakhe ama-agent ahlakaniphile angaxhumana nabantu kanye namanye ama-ejenti. I-LangChain yakhelwe phezu kolimi lohlelo lwe-Rust futhi inikeza i-API elula nenembile yokwakha izinhlelo zokusebenza ze-AI.

Ukuqalisa nge-LangChain

Ukuze uqalise nge-LangChain, uzodinga ukufaka umtapo wezincwadi we-LangChain futhi usethe indawo yokuthuthukisa i-Rust. Nasi isibonelo esilula sendlela yokudala i-ejenti ye-LangChain:

sebenzisa i-langchain::{Agent, Chain};

i-fn main() {
    vumela i-ejenti = I-Agent::new("Umenzeli Wami");
    let chain = Chain::new(ejenti);

    chain.execute("Sawubona, mhlaba!");
}

Le khodi idala i-ejenti entsha ye-LangChain futhi ikhiphe umyalo olula wokuphrinta "Sawubona, mhlaba!" ku-console.

I-DSPy: I-Python-based Agent AI Framework

I-DSPy wuhlaka lwe-AI olusekelwe ku-Python oluhlinzeka nge-API elula nenembile yokwakha ama-agent ahlakaniphile. I-DSPy yakhelwe ukusebenza ngaphandle komthungo nemitapo yolwazi edumile yePython efana neNumPy nePanda.

Ukuqalisa nge-DSPy

Ukuze uqalise nge-DSPy, uzodinga ukufaka umtapo wolwazi we-DSPy futhi usethe indawo yokuthuthukiswa kwePython. Nasi isibonelo esilula sendlela yokudala i-ejenti ye-DSPy:

ngenisa i-dsp njenge-dsp

def main():
    i-ejenti = dsp.Agent("Umenzeli wami")
    i-agent.execute("Sawubona, mhlaba!")

uma __igama__ == "__main__":
    main()

Le khodi idala i-ejenti entsha ye-DSPy futhi yenze umyalo olula wokuphrinta "Sawubona, mhlaba!" ku-console.

Imiyalo/Amazwibela Aphezulu Okuyi-5-10

Nansi eminye yemiyalo evame ukusetshenziswa kakhulu namazwibela we-LangChain ne-DSPy:

LangChain

Umenzeli::omusha("Umenzeli Wami"): Idala umenzeli omusha we-LangChain.
Uchungechunge::okusha(i-ejenti): Idala iketango elisha le-LangChain.
chain.execute("Sawubona, mhlaba!"): Ikhipha umyalo ochungechungeni lwe-LangChain.
i-agent.ask("Ungubani igama lakho?"): Ibuza umenzeli we-LangChain umbuzo.
i-agent.tell("Igama lami nginguJohn."): Utshela umenzeli we-LangChain ucezu lolwazi.

DSPy

dsp.Agent("Umenzeli wami"): Idala i-ejenti entsha ye-DSPy.
i-agent.execute("Sawubona, mhlaba!"): Isebenzisa umyalo kumenzeli we-DSPy.
i-agent.ask("Ungubani igama lakho?"): Ibuza umenzeli we-DSPy umbuzo.
i-agent.tell("Igama lami nginguJohn."): Utshela umenzeli we-DSPy ucezu lolwazi.
i-agent.memory["ukhiye"] = "inani": Isetha inani kumemori yomenzeli we-DSPy.

Ukubona

Nakhu ukubuka ngeso okulula kwe-LangChain nama-ejenti e-DSPy kusetshenziswa i-Chart.js:

{
  "type": "bar",
  "idatha": {
    "labels": ["LangChain", "DSPy"],
    "idathasethi": [
      {
        "label": "Isikhathi Sokwenza Umenzeli",
        "idatha": [10, 20],
        "backgroundColor": ["#4CAF50", "#FF9800"]
      }
    ]
  },
  "izinketho": {
    "isihloko": {
      "Display": iqiniso,
      "text": "Ukuqhathanisa Isikhathi Sokwenza Umenzeli"
    },
    "izikali": {
      "yAxes": [
        {
          "imikhaza": {
            "beginAtZero": kuyiqiniso
          }
        }
      ]
    }
  }
}

Leli shadi liqhathanisa isikhathi sokwenziwa kwe-LangChain nama-ejenti e-DSPy.

Isiphetho

Kulesi sihloko, sihlole ukuthi abaklami be-software nabaphathi bephrojekthi bangasebenzisa kanjani amathuluzi e-AI e-agent afana ne-LangChain ne-DSPy ukuze kuthuthukiswe ukubambisana, izinqubo ezizenzakalelayo, kanye nokuthuthukisa ukulethwa kwesofthiwe ezindaweni zesimanje zokuthuthukiswa. Siphinde sanikeza umhlahlandlela osebenzayo wokuqalisa nge-LangChain ne-DSPy, okuhlanganisa izibonelo ezilula nemiyalo evame ukusetshenziswa namazwibela. Ngokuhlanganisa i-AI ye-agent ekuhambeni komsebenzi wokuthuthukiswa kwesofthiwe, amaqembu angathuthukisa ukusebenza kahle, ukukhiqiza, nokwenza izinqumo, ekugcineni kuholele emiphumeleni engcono yokulethwa kwesofthiwe.

Ukuhlanganisa i-Agentic AI ku-Software Development Workflows: Umhlahlandlela Osebenzayo

¿Tienes preguntas?