Kubatanidza Agentic AI muSoftware Development Workflows: A Practical Guide

Kubatanidza Agentic AI muSoftware Development Workflows: A Inoshanda Guide

Introduction

Iyo software yekuvandudza mamiriro ari kukurumidza kushanduka, nekuwedzera kutorwa kweArtificial Intelligence (AI) uye Machine Kudzidza (ML) matekinoroji. Imwe yenzvimbo dzinonyanya kufadza dzekutsvaga kweAI ndeyeAgentic AI, iyo inotarisa mukugadzira vakangwara vamiririri vanogona kudyidzana nevanhu uye nevamwe vamiririri kuzadzisa zvinangwa zvakaomarara. Muchikamu chino, tichaongorora kuti vanogadzira software uye maneja epurojekiti vanogona sei kukwidziridza maturusi eAI seLangChain neDSPy kusimudzira kubatana, kuita otomatiki maitiro, uye kuvandudza kuendesa software munzvimbo dzemazuva ano dzebudiriro.

Chii chinonzi Agent AI?

Agentic AI inoreva rudzi rweAI iyo yakagadzirirwa kuita pachinzvimbo chevanhu kana vamwe vamiririri kuzadzisa zvinangwa chaizvo. Ava vamiririri vanogona kudyidzana nevanhu, vamwe vamiririri, uye nharaunda kuunganidza ruzivo, kuita sarudzo, uye kutora matanho. Agentic AI yakasiyana neyechinyakare AI mukuti yakagadzirirwa kuve inoshanda uye inozvimiririra, pane kungoita reactive.

Mabhenefiti eAgentic AI muSoftware Development

Saka, nei vagadziri vesoftware uye mamaneja epurojekiti vachifanira kutarisira nezve agent AI? Heano mamwe emabhenefiti ekubatanidza agent AI mune software yekuvandudza workflows:

Kushandirapamwe Kwakavandudzwa: Agentic AI inogona kubatsira zvikwata kuti zvibatane zvinobudirira nekuita mabasa otomatiki, kupa mhinduro yenguva chaiyo, uye kufambisa kutaurirana.
Otomatiki Maitirwo: Agentic AI inokwanisa otomatiki kudzokorora uye inopedza nguva mabasa, ichisunungura vanogadzira kuti vatarise mabasa epamusoro-soro.
Kunatsiridzwa-Kuita Sarudzo: Agentic AI inogona kupa ruzivo rwunofambiswa nedata uye kurudziro kubatsira zvikwata kuita sarudzo dziri nani.
Kuwedzera Kushanda: Agentic AI inogona kubatsira zvikwata kuti zvishande nemazvo nekuona mabhodhoro, kukwenenzvera mafambiro ebasa, uye kugadzirisa maitiro.

**LangChain: Yakakurumbira Agenti AI Framework **

LangChain yakakurumbira agent AI chimiro chinobvumira vanogadzira kuvaka vakangwara vamiririri vanogona kudyidzana nevanhu uye nevamwe vamiririri. LangChain yakavakirwa pamusoro peRust programming mutauro uye inopa iri nyore uye intuitive API yekuvaka agentic AI application.

Kutanga neLangChain

Kuti utange neLangChain, iwe uchafanirwa kuisa raibhurari yeLangChain uye kumisa nzvimbo yekuvandudza Rust. Heino muenzaniso wakapusa wekugadzira mumiriri weLangChain:

shandisa langchain::{Agent, Chain};

fn chikuru() {
    rega mumiririri = Mumiririri :: mutsva ("Mumiririri Wangu");
    regai cheni = Cheni ::nyowani(mumiririri);

    chain.execute("Mhoro, nyika!");
}

Iyi kodhi inogadzira mumiriri mutsva weLangChain uye inoita murairo uri nyore kudhinda "Mhoro, nyika!" ku console.

DSPy: A Python-Yakavakirwa Agentic AI Framework

DSPy iPython-based agent AI framework inopa iri nyore uye intuitive API yekuvaka vakangwara vamiririri. DSPy yakagadzirirwa kushanda zvisina musono nemaraibhurari anozivikanwa ePython seNumPy nePandas.

Kutanga neDSPy

Kuti utange neDSPy, iwe unozofanirwa kuisa iyo DSPy raibhurari uye kuseta Python budiriro nharaunda. Heino muenzaniso wakapfava wekugadzira iyo DSPy mumiriri:

import dSpy se dsp

def chikuru ():
    mumiririri = dsp.Agent("Mumiririri Wangu")
    agent.execute("Mhoro, nyika!")

kana __zita__ == "__main__":
    chikuru()

Iyi kodhi inogadzira itsva DSPy mumiriri uye inoita murairo uri nyore kudhinda "Mhoro, nyika!" ku console.

Pamusoro 5-10 Inonyanya Kushandiswa Mirairo/Zvidimbu

Heano mamwe emirairo anonyanya kushandiswa uye zvimedu zveLangChain uye DSPy:

LangChain

Mumiririri::mutsva("Mumiririri Wangu"): Inogadzira mumiriri mutsva weLangChain.
Chain ::itsva(mumiririri): Inogadzira cheni itsva yeLangChain.
chain.execute("Mhoro, nyika!"): Inoita murairo paLangChain cheni.
agent.ask("Zita rako ndiani?"): Inobvunza mumiririri weLangChain mubvunzo.
agent.tell("Zita rangu ndiJohn."): Inotaurira mumiririri weLangChain chikamu cheruzivo.

DSPy

dsp.Agent("Mumiririri Wangu"): Inogadzira mumiriri mutsva weDSPy.
agent.execute("Mhoro, nyika!"): Inoita murairo pane DSPy mumiririri.
agent.ask("Zita rako ndiani?"): Inobvunza mumiririri weDSPy mubvunzo.
agent.tell("Zita rangu ndiJohn."): Inoudza mumiririri weDSPy chikamu cheruzivo.
agent.memory["kiyi"] = "kukosha": Inoisa kukosha mundangariro yemumiririri weDSPy.

Kuona

Heino ratidziro yakapusa yeLangChain uye DSPy vamiririri vachishandisa Chart.js:

{
  "type": "bha",
  "data": {
    "labels": ["LangChain", "DSPy"],
    "datasets": [
      {
        "label": "Nguva Yekuita Mumiririri",
        "data": [10, 20],
        "backgroundColor": ["#4CAF50", "#FF9800"]
      }
    ]
  },
  "sarudzo": {
    "zita": {
      "kuratidza": chokwadi,
      "zvinyorwa": "Kuenzanisa Kwenguva Yemumiririri"
    },
    "zvikero": {
      "yAxes": [
        {
          "zvikwekwe": {
            "beginAtZero": chokwadi
          }
        }
      ]
    }
  }
}

Iyi chati inofananidza nguva yekuuraya yeLangChain uye DSPy vamiririri.

Mhedziso

Muchinyorwa chino, takaongorora kuti vagadziri vesoftware uye maneja epurojekiti vanogona sei kukwidziridza maajenti eAI maturusi seLangChain uye DSPy kusimudzira kubatana, kuita otomatiki maitiro, uye kugadzirisa software kuunza munzvimbo dzemazuva ano dzebudiriro. Isu takapawo gwara rinoshanda rekutanga neLangChain neDSPy, kusanganisira mienzaniso iri nyore uye mirairo inowanzoshandiswa uye zvimedu. Nekubatanidza agent AI mune software yekuvandudza workflows, zvikwata zvinogona kuvandudza kugona, kugadzira, uye kuita sarudzo, pakupedzisira zvichitungamira kune zvirinani kuburitsa software.

Kubatanidza Agentic AI muSoftware Development Workflows: A Practical Guide

Have Questions?