Leveraging Agentic AI yeSoftware Architecture uye DevOps otomatiki

Nhanganyaya

Iko kukurumidza kushanduka kweArtificial Intelligence (AI) uye Machine Learning (ML) yakatungamirira mukugadzirwa kwe agentic AI, rudzi rweAI inogona kutora matanho uye kuita sarudzo pachayo. Iyi tekinoroji ine mukana wekushandura nzira yatinosvika nayo software yekuvaka uye DevOps otomatiki. Muchikamu chino, tichaongorora kubatanidzwa kweagent AI, Makuru Mutauro Models (LLMs), uye maturusi ekuimba seLangChain neDSPy kuita otomatiki magadzirirwo esoftware, zvivakwa, uye DevOps maitiro.

Chii chinonzi Agent AI?

Agentic AI inoreva rudzi rweAI iyo inogona kutora matanho uye kuita sarudzo pachayo. Izvi zvinopesana nemaitiro echinyakare AI, ayo anowanzo gadzirirwa kuita chaiwo mabasa uye kutevedzera seti yemitemo yakafanotsanangurwa. Agentic AI masisitimu, kune rumwe rutivi, akagadzirwa kuti ave anozvimiririra uye anochinjika, achivabvumira kudzidza kubva kunharaunda yavo uye kuita sarudzo zvichienderana nezvinangwa zvavo uye zvinangwa.

Basa reLLMs muAgentic AI

Makuru Mutauro Models (LLMs) imhando yeAI iyo yakadzidziswa pahuwandu hwemashoko data uye inokwanisa kugadzira mutauro wakafanana nemunhu. MaLLM chinhu chakakosha cheagentic AI masisitimu, sezvo achipa kugona kunzwisisa uye kugadzira mutauro wevanhu. Izvi zvinobvumira agentic AI masisitimu kutaurirana nevanhu uye mamwe masisitimu mune yakasarudzika uye intuitive nzira.

Orchestration Zvishandiso: LangChain uye DSPy

Maturusi eOrchestration akaita seLangChain neDSPy akagadzirirwa kubata nekubatanidza kudyidzana pakati pezvirongwa zvakasiyana uye zvikamu. Zvishandiso izvi zvinopa nzira yekutsanangura nekuita mafambiro ebasa, ayo anogona kusanganisira mabasa akadai sekugadzirisa data, kudzidzisa modhi, uye kutumira.

LangChain

LangChain ndeye yakavhurika-sosi orchestration chishandiso chinopa chinochinjika uye chinowedzera chimiro chekuvaka nekugadzirisa mafambiro ebasa. LangChain yakagadzirirwa kushanda nehuwandu hwehurongwa uye zvikamu, kusanganisira LLMs, databases, uye APIs.

DSPy

DSPy chishandiso chePython-based orchestration chinopa nzira iri nyore uye inonzwisisika yekutsanangura nekuita mafambiro ebasa. DSPy yakagadzirirwa kushanda neyakasiyana siyana masisitimu uye zvikamu, zvinosanganisira LLMs, dhatabhesi, uye APIs.

Automating Software Architecture uye DevOps Maitiro

Agentic AI, LLMs, uye maturusi ekuimba seLangChain neDSPy anogona kushandiswa kugadzira otomatiki akasiyana siyana ekugadzira software uye DevOps maitiro. Mimwe mienzaniso inosanganisira:

Otomatiki kodhi ongororo: Agentic AI masisitimu anogona kushandiswa kuongorora kodhi uye kuona zvingangoitika, senge tsikidzi uye kusachengeteka.
Kuyedza otomatiki: maAgentic AI masisitimu anogona kushandiswa kuita otomatiki kuyedzwa, kusanganisira bvunzo dzeyuniti, kuyedza kubatanidza, uye kupera-kusvika-kumagumo kuyedzwa.
Automated deployment: Agentic AI masisitimu anogona kushandiswa kuita otomatiki kutumira, kusanganisira kutumirwa kwekodhi, kumisikidzwa, uye zvimwe zviwanikwa.
Otomatiki yekutarisisa: maAgentic AI masisitimu anogona kushandiswa kuita otomatiki kutarisa, kusanganisira kuongorora maitiro ehurongwa, kuchengetedza, uye mamwe metrics.

Kushandisa Kushanda

Muchikamu chino, tichapa muenzaniso wakapfava wekushandisa LangChain uye DSPy kuita otomatiki software yekuvaka uye DevOps maitiro.

Muenzaniso: Automating Code Ongororo

Mumuenzaniso uyu, tichashandisa LangChain uye DSPy kuti tiite otomatiki kodhi ongororo. Isu tichatsanangura mafambiro ebasa ayo anosanganisira anotevera matanho:

**Kuongorora kwekodhi **: Kufamba kwebasa kuchashandisa LLM kuongorora kodhi uye kuona zvingangoitika.
** Kuongorora kwekodhi **: Kufamba kwebasa kuchashandisa agentic AI system kuti iongorore kodhi uye kuona zvingangoitika.
** Code deployment **: Basa rekushanda richashandisa chigadziro chekushandisa kutumira kodeti kune imwe nzvimbo yekugadzira.

Heino muenzaniso wekuita uku kufambiswa kwebasa uchishandisa LangChain uye DSPy:

import langchain
kubva ku langchain import LLM

# Tsanangura iyo LLM
llm = LLM("chinyorwa-davinci-003")

# Tsanangura iyo agent AI system
ai_system = langchain.AgenticAI()

# Tsanangura mafambiro ebasa
workflow = langchain.Workflow(
    nhanho=[
        langchain.Step(
            zita = "Code analysis",
            func=llm.analyze_code,
            inputs={"code": "def hello_world(): print('Hello World')"},
        ),
        langchain.Step(
            zita = "Code review",
            func=ai_system.review_code,
            inputs={"code": "def hello_world(): print('Hello World')"},
        ),
        langchain.Step(
            zita = "Code deployment",
            func=langchain.deploy_code,
            inputs={"code": "def hello_world(): print('Hello World')"},
        ),
    ]
)

# Gadzirisa maitiro ekushanda
workflow.execute()

Iyi kodhi inotsanangura mafambiro ebasa anosanganisira matanho matatu: kuongororwa kwekodhi, kuongorora kwekodhi, uye kutumira kwekodhi. Kufambiswa kwebasa kunoshandisa LLM kuongorora kodhi, agentic AI system yekuongorora iyo kodhi, uye chishandiso chekuendesa kuendesa kodhi kunzvimbo yekugadzira.

Yepamusoro 5-10 Yakanyanya Kushandiswa Mirairo / Zvimedu

Heano epamusoro 5-10 anonyanya kushandiswa mirairo/zvimedu zveLangChain uye DSPy:

langchain.Workflow(matanho=[...]): Tsanangura mafambiro ebasa ane matanho akawanda.
langchain.Step(zita="...", func=..., inputs={...}): Tsanangura nhanho imwe chete mukufambiswa kwebasa.
llm.analyze_code(inputs={"code": ...}): Shandisa LLM kuongorora kodhi.
ai_system.review_code(inputs={"code": ...}): Shandisa agentic AI system kuongorora kodhi.
langchain.deploy_code(inputs={"code": ...}): Tumira kodhi kunzvimbo yekugadzira.
langchain.AgenticAI(): Gadzira muenzaniso weiyo agent AI system.
langchain.LLM(): Gadzira muenzaniso weLLM.
langchain.deploy_code(): Tumira kodhi kunzvimbo yekugadzira.
langchain.execute_workflow(): Ita mafambiro ebasa.
langchain.get_workflow_status(): Wana chimiro chekufambiswa kwebasa.

Kuona

Heino muenzaniso wekuona mafambiro ebasa uchishandisa Chart.js:

{
  "type": "bha",
  "data": {
    "labels": ["Code Analysis", "Code Wongororo", "Code Deployment"],
    "datasets": [
      {
        "label": "Matanho ekufambiswa kwebasa",
        "data": [10, 20, 30],
        "backgroundColor": [
          "rgba(255, 99, 132, 0.2)",
          "rgba(54, 162, 235, 0.2)",
          "rgba(255, 206, 86, 0.2)"
        ],
        "borderColor": [
          "rgba(255, 99, 132, 1)",
          "rgba(54, 162, 235, 1)",
          "rgba(255, 206, 86, 1)"
        ],
        "borderWidth": 1
      }
    ]
  },
  "sarudzo": {}
}

Iyi kodhi inotsanangura bhaachati ine mavara matatu: Code Analysis, Code Ongororo, uye Code Deployment. Chati inoratidza nhamba yematanho muchikamu chimwe nechimwe chekufamba kwebasa.

Mhedziso

Agentic AI, LLMs, uye maturusi ekuimba akaita seLangChain neDSPy vane mukana wekushandura nzira yatinosvika nayo magadzirirwo esoftware uye DevOps otomatiki. Nekuita otomatiki mabasa akadai sekuongorora kwekodhi, kuyedzwa, uye kutumira, tinogona kuvandudza mashandiro uye kushanda kwemaitiro edu ebudiriro. Muchinyorwa chino, takaongorora kubatanidzwa kweaya matekinoroji uye takapa muenzaniso wakapfava wekuti ungashandise sei kuita otomatiki software yekuvaka uye DevOps maitiro.