{
  "format": "agent-article-v0",
  "article_id": "EA-086",
  "canonical_url": "/tools/ea-086-serverless-ai-architecture",
  "page_type": "reference",
  "claim_ids": [],
  "evidence_ids": [
    "EVD-EA-086-URL-44B5D8B4F201",
    "EVD-EA-086-URL-C784C5865F16",
    "EVD-EA-086-URL-8CCB457B5EA0",
    "EVD-EA-086-URL-16F9BE23F7EC",
    "EVD-EA-086-URL-728ABC12B871",
    "EA-086-R11-RESEARCH",
    "EA-086-R11-WEB-READBACK"
  ],
  "unknown_ids": [
    "EA-086-UNKNOWN-01",
    "EA-086-UNKNOWN-CURRENTNESS"
  ],
  "operational_object_ids": [],
  "provider": [],
  "models": [],
  "plans": [],
  "surfaces": [],
  "regions": [],
  "current_status": "UNKNOWN",
  "observation_status": "SPEC_ONLY",
  "freshness": "UNKNOWN",
  "last_verified": null,
  "supersedes": [],
  "superseded_by": [],
  "provenance": [
    "COMM-079-EA0086",
    "SRC-EA-086-URL-44B5D8B4F201",
    "SRC-EA-086-URL-C784C5865F16",
    "SRC-EA-086-URL-8CCB457B5EA0",
    "SRC-EA-086-URL-16F9BE23F7EC",
    "SRC-EA-086-URL-728ABC12B871",
    "WEBBODY-EA0086-R8-4F392BAB"
  ],
  "article_sources": [
    {
      "id": "EVD-EA-086-URL-44B5D8B4F201",
      "evidence_class": "RESEARCH",
      "title": "1. official spec/docs | AWS Lambda — Configure Lambda function timeout | Amazon Web Services |",
      "href": "https://docs.aws.amazon.com/lambda/latest/dg/configuration-timeout.html",
      "supports": [],
      "limit": "元の監査済み資料に記載された参照先です。URLの復元は、資料の最新性や全主張の支持を新たに検証したことを意味しません。",
      "origin_class": "SOURCE_PACKET_REFERENCE",
      "source_or_experiment_ids": [
        "SRC-EA-086-URL-44B5D8B4F201"
      ],
      "supporting_locators": [
        "EVIDENCE TABLE",
        "1. official spec/docs | AWS Lambda — Configure Lambda function timeout | Amazon Web Services | https://docs.aws.amazon.com/lambda/latest/dg/configuration-timeout.html"
      ],
      "scope": "EVIDENCE TABLE",
      "provenance_status": "SOURCE_PACKET_REFERENCE_PROJECTED"
    },
    {
      "id": "EVD-EA-086-URL-C784C5865F16",
      "evidence_class": "RESEARCH",
      "title": "2. official spec/docs | Using Lambda with Amazon SQS | Amazon Web Services |",
      "href": "https://docs.aws.amazon.com/lambda/latest/dg/with-sqs.html",
      "supports": [],
      "limit": "元の監査済み資料に記載された参照先です。URLの復元は、資料の最新性や全主張の支持を新たに検証したことを意味しません。",
      "origin_class": "SOURCE_PACKET_REFERENCE",
      "source_or_experiment_ids": [
        "SRC-EA-086-URL-C784C5865F16"
      ],
      "supporting_locators": [
        "EVIDENCE TABLE",
        "2. official spec/docs | Using Lambda with Amazon SQS | Amazon Web Services | https://docs.aws.amazon.com/lambda/latest/dg/with-sqs.html"
      ],
      "scope": "EVIDENCE TABLE",
      "provenance_status": "SOURCE_PACKET_REFERENCE_PROJECTED"
    },
    {
      "id": "EVD-EA-086-URL-8CCB457B5EA0",
      "evidence_class": "RESEARCH",
      "title": "3. official spec/docs | Cloud Run GPU support for services | Google Cloud |",
      "href": "https://docs.cloud.google.com/run/docs/configuring/services/gpu",
      "supports": [],
      "limit": "元の監査済み資料に記載された参照先です。URLの復元は、資料の最新性や全主張の支持を新たに検証したことを意味しません。",
      "origin_class": "SOURCE_PACKET_REFERENCE",
      "source_or_experiment_ids": [
        "SRC-EA-086-URL-8CCB457B5EA0"
      ],
      "supporting_locators": [
        "EVIDENCE TABLE",
        "3. official spec/docs | Cloud Run GPU support for services | Google Cloud | https://docs.cloud.google.com/run/docs/configuring/services/gpu"
      ],
      "scope": "EVIDENCE TABLE",
      "provenance_status": "SOURCE_PACKET_REFERENCE_PROJECTED"
    },
    {
      "id": "EVD-EA-086-URL-16F9BE23F7EC",
      "evidence_class": "RESEARCH",
      "title": "4. official spec/docs | Create jobs — Cloud Run | Google Cloud |",
      "href": "https://cloud.google.com/run/docs/create-jobs",
      "supports": [],
      "limit": "元の監査済み資料に記載された参照先です。URLの復元は、資料の最新性や全主張の支持を新たに検証したことを意味しません。",
      "origin_class": "SOURCE_PACKET_REFERENCE",
      "source_or_experiment_ids": [
        "SRC-EA-086-URL-16F9BE23F7EC"
      ],
      "supporting_locators": [
        "EVIDENCE TABLE",
        "4. official spec/docs | Create jobs — Cloud Run | Google Cloud | https://cloud.google.com/run/docs/create-jobs"
      ],
      "scope": "EVIDENCE TABLE",
      "provenance_status": "SOURCE_PACKET_REFERENCE_PROJECTED"
    },
    {
      "id": "EVD-EA-086-URL-728ABC12B871",
      "evidence_class": "RESEARCH",
      "title": "5. official spec/docs | Azure Functions Scale and Hosting | Microsoft |",
      "href": "https://learn.microsoft.com/azure/azure-functions/functions-scale",
      "supports": [],
      "limit": "元の監査済み資料に記載された参照先です。URLの復元は、資料の最新性や全主張の支持を新たに検証したことを意味しません。",
      "origin_class": "SOURCE_PACKET_REFERENCE",
      "source_or_experiment_ids": [
        "SRC-EA-086-URL-728ABC12B871"
      ],
      "supporting_locators": [
        "EVIDENCE TABLE",
        "5. official spec/docs | Azure Functions Scale and Hosting | Microsoft | https://learn.microsoft.com/azure/azure-functions/functions-scale"
      ],
      "scope": "EVIDENCE TABLE",
      "provenance_status": "SOURCE_PACKET_REFERENCE_PROJECTED"
    },
    {
      "id": "EA-086-R11-RESEARCH",
      "evidence_class": "RESEARCH",
      "title": "EA-086 監査済みresearch / canonical revision binding",
      "supports": [
        "1. Serverless AIの適切なCoreは「LLMをfunction内へ必ず載せる」ではない。小さなorchestrator/functionがeventを受け、durable queue/stateへ記録し、managed model APIまたは別inference serviceを呼び、結果をdurable storeへcommitする形が最もportableである。\n2. execution timeoutはprovider/plan/runtimeで異なる。AWS Lambda標準functionは最大15分、Azure Functionsはhosting planで大きく異なりHTTP応答には別の230秒境界、Cloud Run Jobsは通常task最大7日だがGPU taskは1時間である。よって「serverlessだから短時間」も「container serverlessなら無制限」も一般化不可。\n3. queue/event integrationはduplicateを前提にする。AWS Lambda+SQSはat-least-onceでduplicate processingを明記しているため、AI call・billing・email送信・DB mutation等のside effectはidempotency key、deduplication record、outbox/transaction boundaryが必要。\n4. GPU serverlessは実在するがavailability gateが強い。Cloud Run GPUはscale-to-zero可能で、L4/RTX PRO 6000 Blackwellをmanagedで提供する一方、GPU type、region、CPU/RAM minimum、one-GPU-per-instance等の制約がある。GPU availabilityをCPU serverlessと同等に扱わない。\n5. cold startだけでなくmodel initialization、weight download、tokenizer/runtime initialization、network path、upstream model rate limitを別々に測る必要がある。platform instance startup値をend-to-end TTFT保証へ外挿しない。"
      ],
      "limit": "canonical監査・revision bindingの存在を示す。外部ソースcurrentnessが本文またはcontrol-planeで確定していない場合は、currentnessをUNKNOWNのまま保持する。",
      "origin_class": "AUDITED_CANONICAL_RESEARCH_PACKET",
      "source_or_experiment_ids": [
        "COMM-079-EA0086"
      ],
      "supporting_locators": [
        "COMM-079-EA0086",
        "EA-086-R5-LEGACY-MIGRATION-20261008-R1",
        "content/staging/EA-086.md"
      ],
      "scope": "監査済みEA本文・canonical revision・既存Source Graph relationの範囲。欠損currentnessは推測で閉じない。",
      "provenance_status": "AUDIT_PASS_CANONICAL_BINDING"
    },
    {
      "id": "EA-086-R11-WEB-READBACK",
      "evidence_class": "IMPL",
      "title": "EA-086 Web body identity / canonical readback",
      "supports": [
        "canonical staging bodyのSHA-256一致とR9 canonical ledger readbackをProduction projectionの入力として検証する。"
      ],
      "limit": "body identity/readbackは本文claim自体の真偽や外部ソースの最新性を追加保証しない。",
      "checked_at": null,
      "origin_class": "IMPLEMENTATION_READBACK",
      "source_or_experiment_ids": [
        "WEBBODY-EA0086-R8-4F392BAB"
      ],
      "supporting_locators": [
        "WEBBODY-EA0086-R8-4F392BAB",
        "content/staging/EA-086.md",
        "EA-086-R5-LEGACY-MIGRATION-20261008-R1"
      ],
      "scope": "R9/R10 canonical readback後のProduction生成時body identity。",
      "provenance_status": "CANONICAL_READBACK_BODY_HASH_MATCH"
    }
  ],
  "research_sections": [
    {
      "heading": "WHY_THIS_MATTERS_TO_UPDATE",
      "text": "小規模AI業務では常時稼働clusterを持たず、event-triggered function、queue、managed container、scale-to-zero inferenceを組み合わせることで運用負荷とidle costを下げられる。一方「serverless」は無制限・状態保持・exactly-once・GPU常時即応を意味しない。AI API latency、cold start、model load、retry、duplicate delivery、execution timeout、GPU region/shape availabilityを設計契約へ明示しないと、安価なPoCから本番へ移る際にfailure envelopeが急拡大する。"
    },
    {
      "heading": "SCOPE_ROUTING",
      "text": "Enterprise AI Core / infrastructure / implementation / interoperability / governance / cost / operability / provider mapping"
    },
    {
      "heading": "CLAIM_OR_RESEARCH_QUESTION",
      "text": "Serverless AIは、AI modelそのものではなく「短命compute + durable queue/state + idempotent handler + external/managed model endpoint」を基本形にするとboundedに成立するか。GPU serverless/container serverlessを含め、どの制約をavailability gateとして分離すべきか。"
    },
    {
      "heading": "EVIDENCE TABLE",
      "text": "1. official spec/docs | AWS Lambda — Configure Lambda function timeout | Amazon Web Services | https://docs.aws.amazon.com/lambda/latest/dg/configuration-timeout.html | accessed 2026-09-16 | current docs | locator: timeout default 3s; standard function max 900s/15min; Managed Instances async/event-source mappingの一部は5400s/90min | VERIFIED_SPEC / PRODUCT_VOLATILE | support + limit\n2. official spec/docs | Using Lambda with Amazon SQS | Amazon Web Services | https://docs.aws.amazon.com/lambda/latest/dg/with-sqs.html | accessed 2026-09-16 | current docs | locator: event source mapping is at-least-once; duplicate processing can occur; idempotent function recommended | VERIFIED_SPEC | support + counter\n3. official spec/docs | Cloud Run GPU support for services | Google Cloud | https://docs.cloud.google.com/run/docs/configuring/services/gpu | accessed 2026-09-16 | current docs | locator: RTX PRO 6000 Blackwell 96GB / L4 24GB; one GPU per instance; scale-to-zero; on-demand; GPU-equipped instance startup approx. 5s; supported regions enumerated | VERIFIED_SPEC / PRODUCT_VOLATILE | support + availability limit\n4. official spec/docs | Create jobs — Cloud Run | Google Cloud | https://cloud.google.com/run/docs/create-jobs | accessed 2026-09-16 | current docs | locator: task timeout default 10min, configurable to 168h; GPU task max 1h; retry configurable | VERIFIED_SPEC / PRODUCT_VOLATILE | support + limit\n5. official spec/docs | Azure Functions Scale and Hosting | Microsoft | https://learn.microsoft.com/azure/azure-functions/functions-scale | accessed 2026-09-16 | current docs | locator: Flex/Premium/Dedicated/Container Apps timeout differs; Consumption default 5min/max10min; HTTP-triggered response max 230s due load balancer; async pattern recommended for longer work | VERIFIED_SPEC / PRODUCT_VOLATILE | context + limit"
    },
    {
      "heading": "FINDINGS",
      "text": "1. Serverless AIの適切なCoreは「LLMをfunction内へ必ず載せる」ではない。小さなorchestrator/functionがeventを受け、durable queue/stateへ記録し、managed model APIまたは別inference serviceを呼び、結果をdurable storeへcommitする形が最もportableである。\n2. execution timeoutはprovider/plan/runtimeで異なる。AWS Lambda標準functionは最大15分、Azure Functionsはhosting planで大きく異なりHTTP応答には別の230秒境界、Cloud Run Jobsは通常task最大7日だがGPU taskは1時間である。よって「serverlessだから短時間」も「container serverlessなら無制限」も一般化不可。\n3. queue/event integrationはduplicateを前提にする。AWS Lambda+SQSはat-least-onceでduplicate processingを明記しているため、AI call・billing・email送信・DB mutation等のside effectはidempotency key、deduplication record、outbox/transaction boundaryが必要。\n4. GPU serverlessは実在するがavailability gateが強い。Cloud Run GPUはscale-to-zero可能で、L4/RTX PRO 6000 Blackwellをmanagedで提供する一方、GPU type、region、CPU/RAM minimum、one-GPU-per-instance等の制約がある。GPU availabilityをCPU serverlessと同等に扱わない。\n5. cold startだけでなくmodel initialization、weight download、tokenizer/runtime initialization、network path、upstream model rate limitを別々に測る必要がある。platform instance startup値をend-to-end TTFT保証へ外挿しない。"
    },
    {
      "heading": "LIMITS / COUNTEREVIDENCE / FAILURE ENVELOPE",
      "text": "- function timeout前にupstream model callが長引けばpartial work/duplicate retryが起こり得る。\n- queue redeliveryは障害ではなくdelivery semanticsの一部。handler非idempotentなら二重課金・二重通知・二重mutationになり得る。\n- scale-to-zeroはidle costを下げる一方、cold path latencyを増やし得る。\n- GPU startup約5秒というCloud Run値はcontainer内model readiness/weight load/TTFTを保証しない。\n- serverless providerのquota、concurrency、region、GPU inventory、network egress、secret/IAM設定はtenantごとに異なる。\n- long-running agentを単一request/function invocationへ閉じ込めるとtimeout・retry・checkpoint lossが増える。durable workflow/state machineへ分解すべきケースがある。"
    },
    {
      "heading": "AVAILABILITY_GATES",
      "text": "provider / region / hosting plan / runtime generation / trigger type / timeout / concurrency quota / queue semantics / GPU type and inventory / CPU-memory minimum / VPC-network path / model endpoint region / secret manager and IAM / payload size / retry-DLQ policy / pricing revision。"
    },
    {
      "heading": "IMPLEMENTATION_STATE",
      "text": "VERIFIED_SPEC: Lambda+SQS、Cloud Run services/jobs/GPU、Azure Functionsで実装可能な構成要素を確認。PROPOSAL: 将来Coreに `SERVERLESS_AI_EXECUTION_CONTRACT` を置く。fields: event_id/idempotency_key, trigger/source, queue/topic, payload hash/schema, function/container image digest, runtime revision, timeout, retry/backoff, DLQ, model endpoint/model ID, request hash, response artifact, durable checkpoint, side-effect commit key, region, concurrency, cold/warm marker, latency breakdown, cost attribution, audit trace。"
    },
    {
      "heading": "BENCHMARK_OR_EVAL",
      "text": "TEST_PROFILE（未実施）: warm/cold、single/burst、duplicate event、upstream 429/5xx、timeout直前response、poison message、partial batch failure、DLQ replay、model endpoint latency spike、GPU scale-from-zeroを固定fixture化。p50/p95/p99 end-to-end latency、cold-start contribution、duplicate side-effect rate、retry amplification、DLQ rate、success after retry、cost/request、idle cost、model-ready time、audit completenessを分離計測する。"
    },
    {
      "heading": "SECURITY / PRIVACY / AUDIT / COST / OPERABILITY",
      "text": "Security: least-privilege service identity、secret manager、private egress/VPC要否、webhook/event authenticity、payload classificationを別管理。Privacy: queue/log/DLQにpromptやPIIが残る可能性をretention対象化。Audit: event→invocation→model request→side effectをtrace/idempotency keyで結ぶ。Cost: function/container computeだけでなくqueue、egress、model tokens、GPU startup/idle、retry amplificationを含める。Operability: DLQ replay、rollback、quota exhaustion、region failoverをrunbook化。"
    },
    {
      "heading": "SAFE_WORDING_FOR_FUTURE_BOOK",
      "text": "「Serverless AIは、小規模・断続的・event-drivenなAI処理で常時稼働基盤を減らせる有力な実装形である。ただし実行時間、再試行、重複配送、quota、GPU/region、cold pathはproviderごとに異なる。外部side effectを伴う処理ではdurable stateとidempotencyを設計契約に含める。」"
    },
    {
      "heading": "UPDATE_WORK_HANDOFF",
      "text": "1. provider mapping時にAWS Lambda/SQS、Google Cloud Run/Tasks/Pub/Sub、Azure Functions/Service Bus等を同一比較軸へmapする。\n2. pricing・quota・GPU region/typeは公開後更新時に再pinする。\n3. EA-085 Event-driven AIのEVENT_TO_ACTION_CONTRACTと重複統合せず、EA-086はexecution substrate/cost/timeout/cold pathを担当する。\n4. long-running agent/workflowはDurable Functions/Step Functions/Workflows等のdurable orchestrationを別subsectionで追加採集する。"
    },
    {
      "heading": "OPEN_UNKNOWN",
      "text": "2026-09-16時点の各provider exact price/concurrency quota、Cloud Run GPUの全region inventory、Azure Flex Consumptionのtenant別availability、AWS Lambda Managed Instancesの全適用条件。STOP_CONDITION: execution semantics、duplicate/retry boundary、GPU serverless availability gate、safe architecture contractが一次資料で確立したためEA-086の初回bounded packetは完了。価格/region/quotaは公開後Workで再確認する。\n\n\n---\n\n---"
    }
  ],
  "human_url": "/tools/ea-086-serverless-ai-architecture",
  "markdown_url": "/agent/article.md?path=%2Ftools%2Fea-086-serverless-ai-architecture",
  "json_url": "/agent/article.json?path=%2Ftools%2Fea-086-serverless-ai-architecture"
}