Leveraging Agentic AI mo Software Architecture ma DevOps Automation

Folasaga

O le televave o le atinaʻeina o le Artificial Intelligence (AI) ma le Machine Learning (ML) ua taʻitaʻia ai le atinaʻeina o le AI faʻapitoa, o se ituaiga AI e mafai ona faia ni gaioiga ma faia filifiliga na o ia. O lenei tekinolosi e iai le gafatia e toe faʻafouina ai le auala tatou te faʻalatalata atu ai i le faʻataʻitaʻiga o masini komepiuta ma le masini DevOps. I totonu o lenei tusiga, o le a matou suʻesuʻeina le tuʻufaʻatasia o AI sui, Large Language Models (LLMs), ma meafaigaluega faʻapipiʻi e pei o LangChain ma DSPy e faʻatometi ai le fausaga faʻapipiʻi, atinaʻe, ma faiga DevOps.

O le a le Agentic AI?

Agentic AI e faasino i se ituaiga AI e mafai ona faia ni gaioiga ma faia filifiliga na o ia. E faʻatusatusa lea i faiga masani a AI, lea e masani ona mamanuina e faʻatino ai galuega faʻapitoa ma mulimuli i se seti o tulafono faʻatulagaina. Agentic AI systems, i le isi itu, ua mamanuina ina ia sili atu le tutoʻatasi ma fetuutuunai, faʻatagaina i latou e aʻoaʻo mai lo latou siosiomaga ma faia filifiliga e faʻatatau i a latou lava sini ma sini.

Le Matafaioi a LLM i Agentic AI

Large Language Models (LLMs) o se ituaiga AI o loʻo aʻoaʻoina i le tele o faʻamaumauga o tusitusiga ma e mafai ona faʻatupuina gagana pei o tagata. LLMs o se vaega autu o faiga faʻapitoa AI, aua latou te maua le tomai e malamalama ai ma faʻatupuina le gagana a le tagata. Ole mea lea e mafai ai e faʻalapotopotoga faʻapitoa AI ona fesoʻotaʻi ma tagata ma isi faiga i se auala sili atu ona masani ma faʻaogaina.

Meafaigaluega: LangChain ma DSPy

O mea faigaluega faʻapipiʻi e pei o LangChain ma DSPy ua mamanuina e pulea ma faʻamaopoopo fegalegaleaiga i le va o faiga eseese ma vaega. O nei meafaigaluega e tuʻuina atu ai se auala e faʻamalamalamaina ma faʻatino galuega, lea e mafai ona aofia ai galuega e pei o faʻamaumauga faʻamaumauga, aʻoaʻoga faʻataʻitaʻiga, ma le faʻapipiʻiina.

LangChain

LangChain o se meafaigaluega faʻapipiʻi punaʻoa tatala lea e maua ai se faʻatulagaga fetuutuunai ma faʻalauteleina mo le fausiaina ma le puleaina o galuega. LangChain ua mamanuina e galue ai ma le tele o faiga ma vaega, e aofia ai LLM, faʻamaumauga, ma API.

DSPy

DSPy o se meafaigaluega faʻapipiʻi faʻavae Python e maua ai se auala faigofie ma faʻaogaina e faʻamalamalamaina ma faʻatino galuega. DSPy ua mamanuina e galue ma le tele o faiga ma vaega, e aofia ai LLMs, databases, ma APIs.

Otometi Faiga Fa'ata'ita'iga ma DevOps

Agentic AI, LLMs, ma meafaigaluega faʻapipiʻi e pei o LangChain ma DSPy e mafai ona faʻaogaina e faʻaogaina ai le tele o faʻasologa o polokalama faakomepiuta ma faiga DevOps. O nisi fa'ata'ita'iga e aofia ai:

Su'esu'eina tulafono fa'apitoa: E mafai ona fa'aogaina faiga fa'apitoa AI e toe iloilo ai fa'ailoga ma fa'ailoa ai fa'afitauli e ono tutupu mai, e pei o pusi ma fa'aletonu saogalemu.
Su'ega otometi: E mafai ona fa'aogaina faiga fa'apitoa AI e otometi ai su'ega, e aofia ai su'ega iunite, su'ega tu'ufa'atasi, ma su'ega fa'ai'u.
Otometi fa'apipi'i: Agentic AI e mafai ona fa'aogaina e otometi ai le fa'apipi'iina, e aofia ai le fa'apipi'iina o code, fa'atulagaina, ma isi punaoa.
Otometi mata'ituina: Agentic AI e mafai ona fa'aogaina e otometi ai le mata'ituina, e aofia ai le mata'ituina o le fa'atinoga o faiga, saogalemu, ma isi fua.

Fa'atinoga Fa'atino

I lenei vaega, o le a matou tuʻuina atu se faʻataʻitaʻiga faigofie o le faʻaogaina o le LangChain ma le DSPy e faʻatometi ai se faʻataʻitaʻiga polokalama ma le DevOps process.

Fa'ata'ita'iga: Otometi Fa'amatalaga Fa'ailoga

I lenei faʻataʻitaʻiga, o le a matou faʻaogaina LangChain ma DSPy e faʻatautaia le iloiloga o tulafono. O le a matou faʻamatalaina se faʻasologa o galuega e aofia ai laasaga nei:

Su'esu'ega o tulafono: O le a fa'aogaina e le fa'asologa o galuega le LLM e fa'avasega ai le fa'ailoga ma fa'ailoa ai fa'afitauli e ono tutupu mai.
Code review: O le a fa'aogaina e le fa'asologa o galuega se faiga fa'apitoa AI e toe iloilo ai le tulafono ma fa'ailoa ai fa'afitauli e ono tula'i mai.
Code deployment: Ole galuega ole a fa'aogaina se meafaigaluega fa'apipi'i e fa'apipi'i ai le code ile si'osi'omaga gaosiga.

O se faʻataʻitaʻiga lea o le faʻaogaina o lenei galuega faʻaaoga LangChain ma DSPy:

faaulufale mai le langchain
mai langchain faaulufale mai LLM

# Fa'amatala le LLM
llm = LLM("text-davinci-003")

# Fa'amatala le faiga ole AI
ai_system = langchain.AgenticAI()

# Fa'amatala le fa'asologa o galuega
galuega = langchain.Workflow(
    laasaga=[
        langchain.Laasaga(
            igoa = "Code analysis",
            func=llm.analyze_code,
            inputs={"code": "def hello_world(): lolomi('Talofa le Lalolagi')"},
        ),
        langchain.Laasaga(
            igoa = "Code review",
            func=ai_system.review_code,
            inputs={"code": "def hello_world(): lolomi('Talofa le Lalolagi')"},
        ),
        langchain.Laasaga(
            igoa = "Code deployment",
            func=langchain.deploy_code,
            inputs={"code": "def hello_world(): lolomi('Talofa le Lalolagi')"},
        ),
    ]
)

# Faʻatino le faʻasologa o galuega
workflow.execute()

O lenei tulafono e faʻamatalaina ai se faʻasologa o galuega e aofia ai laasaga e tolu: suʻesuʻega tulafono, iloiloga tulafono, ma le faʻaogaina o tulafono. E fa'aogaina e le fa'asologa o galuega le LLM e fa'avasega ai le fa'ailoga, se faiga fa'apitoa AI e toe iloilo ai le tulafono, ma se meafaigaluega fa'apipi'i e fa'apipi'i ai le code i se siosiomaga gaosiga.

Top 5-10 Poloaiga/Snippets Tele Fa'aaogaina

O le pito i luga 5-10 pito sili ona faʻaaogaina poloaiga / snippets mo LangChain ma DSPy:

langchain.Workflow(step=[...]): Fa'amatala se fa'asologa o galuega ma ni laasaga se tele.
langchain.Step(name="...", func=..., inputs={...}): Fa'amatala se la'asaga e tasi i le fa'asologa o galuega.
llm.analyze_code(inputs={"code": ...}): Fa'aoga se LLM e iloilo ai le code.
ai_system.review_code(inputs={"code": ...}): Fa'aoga se faiga fa'apitoa AI e toe iloilo ai le code.
langchain.deploy_code(inputs={"code": ...}): Tu'u le code i se siosiomaga gaosiga.
langchain.AgenticAI(): Fausia se faʻataʻitaʻiga o se faiga faʻapitoa AI.
langchain.LLM(): Fausia se fa'ata'ita'iga o se LLM.
langchain.deploy_code(): Tu'u le code i se siosiomaga gaosiga.
langchain.execute_workflow(): Fa'atino se galuega.
langchain.get_workflow_status(): Maua le tulaga o le galuega.

Va'aiga

O se fa'ata'ita'iga lea o le va'ava'ai i se fa'asologa o galuega e fa'aaoga ai le Chart.js:

{
  "type": "pa",
  "fa'amatalaga": {
    "labels": ["Code Analysis", "Code Review", "Code Deployment"],
    "fa'amaumauga": [
      {
        "label": "Laasaga o Galuega",
        "faamatalaga": [10, 20, 30],
        "Lulu lanu": [
          "rgba(255, 99, 132, 0.2)",
          "rgba(54, 162, 235, 0.2)",
          "rgba(255, 206, 86, 0.2)"
        ],
        "Lulu tuaoi": [
          "rgba(255, 99, 132, 1)",
          "rgba(54, 162, 235, 1)",
          "rgba(255, 206, 86, 1)"
        ],
        "BorderWidth": 1
      }
    ]
  },
  "filifiliga": {}
}

O lenei fa'ailoga o lo'o fa'amatalaina ai se siata pa'u ma fa'ailoga e tolu: Su'esu'ega Fa'asinomaga, Iloiloga Fa'asinomaga, ma Fa'asoa Fa'ailoga. O le siata o loʻo faʻaalia ai le numera o laʻasaga i laʻasaga taʻitasi o le galuega.

Fa'ai'uga

Agentic AI, LLMs, ma meafaigaluega faʻapipiʻi e pei o LangChain ma DSPy e iai le avanoa e faʻafouina ai le auala tatou te faʻaogaina ai le faʻaogaina o masini komepiuta ma le masini DevOps. E ala i le otometi galuega e pei o le iloiloga code, su'ega, ma le fa'atinoina, e mafai ona fa'aleleia atili le lelei ma le lelei o a tatou faiga fa'atupuina. I totonu o lenei tusiga, ua matou suʻesuʻeina le tuʻufaʻatasia o nei tekinolosi ma tuʻuina atu se faʻataʻitaʻiga faigofie o le faʻaogaina o ia mea e faʻaogaina ai se faʻataʻitaʻiga polokalama ma le DevOps process.