Knowledge Base as an Engine: Built for RAG, Not for Reading

{
  "id": "ams-020",
  "title": "Knowledge Base as an Engine: Built for RAG, Not for Reading",
  "hook": "A classic SAP KB is dead text. A modern AMS knowledge base is an execution engine for humans, bots, and AI.",
  "idea": "If knowledge can’t be retrieved precisely under pressure, it doesn’t exist. AMS knowledge must be structured, opinionated, and machine-readable from day one.",
  "sap_specific_problem": {
    "what_fails_today": [
      "Long wiki pages without decision points",
      "Search by keywords instead of symptoms",
      "Knowledge written once and never validated",
      "Docs that explain SAP, not how *your* system behaves"
    ]
  },
  "rag_first_principles": [
    "Small atomic chunks beat long explanations.",
    "Structure beats prose.",
    "Decisions beat descriptions.",
    "Examples beat theory."
  ],
  "knowledge_atoms": {
    "incident_atom": [
      "symptom_pattern",
      "business_impact_type",
      "systems_and_clients",
      "objects_involved",
      "first_3_checks",
      "likely_root_causes",
      "fix_options_ranked",
      "rollback_or_workaround",
      "prevention_notes",
      "evidence_examples"
    ],
    "change_atom": [
      "change_type",
      "blast_radius",
      "pre_checks",
      "test_cases",
      "verification_steps",
      "rollback_plan",
      "known_side_effects"
    ],
    "decision_atom": [
      "situation",
      "options",
      "tradeoffs",
      "recommended_choice",
      "when_not_to_do_it"
    ]
  },
  "sap_taxonomy": {
    "dimensions": [
      "Business flow (OTC, P2P, RTR, MDM)",
      "Failure mode (data, auth, integration, logic, performance)",
      "Object type (BP, material, order, IDoc, job)",
      "Risk level (low/medium/high)",
      "Change sensitivity (stable/volatile)"
    ],
    "rule": "Every knowledge atom must be tagged across at least 3 dimensions."
  },
  "knowledge_lifecycle": {
    "create": "Born from real incidents, changes, and RCAs — never hypothetical.",
    "validate": "Used at least once in a real case or drill.",
    "measure": "Track whether it reduced diagnosis or resolution time.",
    "retire": "Delete or rewrite if it’s unused or misleading."
  },
  "automation": {
    "copilot_moves": [
      "Convert incident timelines into knowledge atoms automatically.",
      "Suggest missing atoms for high-incident areas.",
      "Answer chat questions using KB + live context (RAG).",
      "Flag stale or contradictory knowledge."
    ],
    "human_role": [
      "Approve and refine atoms.",
      "Decide what is ‘blessed truth’ vs historical note."
    ]
  },
  "why_this_changes_ams": [
    "L1/L2 become effective without guessing.",
    "Experts stop answering the same questions.",
    "AI assistance becomes reliable instead of hallucinating."
  ],
  "metrics_that_keep_it_sharp": [
    "RAG answer success rate",
    "KB usage during P1/P2 incidents",
    "Time-to-first-hypothesis reduction",
    "Stale atom ratio"
  ],
  "anti_patterns_to_kill": [
    "KB as a dumping ground",
    "Writing docs after memory fades",
    "One-size-fits-all SAP explanations"
  ],
  "design_question": [
    "Can the system answer the question faster than the senior consultant?"
  ],
  "meta": {
    "schema": "dkharlanau.dataset.byte",
    "schema_version": "1.1",
    "dataset": "ams",
    "source_project": "cv-ai",
    "source_path": "ams/ams-020.json",
    "generated_at_utc": "2026-02-03T14:33:32+00:00",
    "creator": {
      "name": "Dzmitryi Kharlanau",
      "role": "SAP Lead",
      "website": "https://dkharlanau.github.io",
      "linkedin": "https://www.linkedin.com/in/dkharlanau"
    },
    "attribution": {
      "attribution_required": true,
      "preferred_citation": "Dzmitryi Kharlanau (SAP Lead). Dataset bytes: https://dkharlanau.github.io"
    },
    "license": {
      "name": "",
      "spdx": "",
      "url": ""
    },
    "links": {
      "website": "https://dkharlanau.github.io",
      "linkedin": "https://www.linkedin.com/in/dkharlanau"
    },
    "contact": {
      "preferred": "linkedin",
      "linkedin": "https://www.linkedin.com/in/dkharlanau"
    },
    "canonical_url": "https://dkharlanau.github.io/datasets/ams/ams-020.json",
    "created_at_utc": "2026-02-03T14:33:32+00:00",
    "updated_at_utc": "2026-02-03T15:29:02+00:00",
    "provenance": {
      "source_type": "chat_export_extraction",
      "note": "Extracted and curated by Dzmitryi Kharlanau; enriched for attribution and crawler indexing."
    },
    "entity_type": "ams_byte",
    "entity_subtype": "",
    "summary": "A classic SAP KB is dead text. A modern AMS knowledge base is an execution engine for humans, bots, and AI."
  }
}