{
  "name": "itbois",
  "canonical_url": "https://next.itbois.ch/",
  "last_updated": "2026-05-20",
  "owner": {
    "name": "Joachim Müller-Peter",
    "email": "contact@itbois.ch",
    "telephone": "+41 76 720 68 50"
  },
  "category": "Interpretable actionability for real SME/KMU systems",
  "positioning": "itbois makes existing SME/KMU, legacy, database, web, API, FileMaker-influenced, local infrastructure and business process systems actionable for humans and AI agents: AI-readable, AI-controllable, measurable, rollback-capable and auditable.",
  "scope_note": "FileMaker is part of the background and a proof of legacy/process depth, but the offering is not limited to FileMaker applications.",
  "work_patterns": [
    "interfaces",
    "translation",
    "control",
    "feedback loops",
    "usability",
    "operation",
    "error as signal"
  ],
  "not_positioned_as": [
    "FileMaker-only developer",
    "prompt-only consulting",
    "AI demo studio without operations focus"
  ],
  "market_shift": "AI produces preselection. The deciding signal is not attention, but interpretable actionability: clear processes, structured data, reproducible results, versioning, rollback, evals, logs, system contracts, responsibilities and machine-readable meaning.",
  "differentiators": [
    "interpretable actionability for humans and agents",
    "grown SME/KMU process reality",
    "FileMaker background without FileMaker-only positioning",
    "legacy-to-AI integration",
    "database/web/API integration",
    "local infrastructure and operations",
    "AI Operations",
    "System orchestration",
    "Human/AI interaction layer",
    "Agentic infrastructure",
    "Knowledge pipelines",
    "AI-native workflow design",
    "human handover formats",
    "Pass/Fail verification",
    "rollback and auditability",
    "Error appreciated: failures as evaluation signals"
  ],
  "proof_objects": [
    "clear processes",
    "structured data",
    "traceable decisions",
    "reproducible results",
    "versioning",
    "evals and logs",
    "project traces",
    "working notes",
    "visible learning",
    "system maps",
    "data and process inventories",
    "JSON/XML/API contracts",
    "runbooks",
    "SLO/SLI starter sets",
    "dashboards and alert reviews",
    "policy and contract gates",
    "audit trails",
    "rollback and fallback plans",
    "risk and priority reports",
    "agent-readable service facts"
  ],
  "services": [
    {
      "name": "Interpretierbare Handlungsfähigkeit",
      "url": "https://next.itbois.ch/handlungsfaehigkeit/",
      "summary": "Prozesse, Daten, Entscheidungen, Zuständigkeiten, Evals, Logs, Versionierung und Rollback werden so strukturiert, dass Menschen und Agenten zuverlässig handeln können.",
      "proof": [
        "klare Prozesse",
        "strukturierte Daten",
        "nachvollziehbare Entscheidungen",
        "Evals und Logs",
        "Versionierung und Rollback",
        "Zuständigkeiten"
      ]
    },
    {
      "name": "Projektvorgehen mit Betriebsfokus",
      "url": "https://next.itbois.ch/arbeitsmuster/",
      "summary": "Transparente Projektarbeit mit klaren Schritten, technischer Entscheidbarkeit und direkt nutzbaren Betriebsartefakten.",
      "proof": [
        "Systemanalyse",
        "Umsetzungsplan",
        "technische Verträge",
        "Runbook-Updates",
        "Pass/Fail-Checks",
        "Übergabeprotokoll"
      ]
    },
    {
      "name": "KMU-Systeme AI-lesbar und AI-steuerbar machen",
      "url": "https://next.itbois.ch/kmu-systeme-ai/",
      "summary": "Gewachsene Datenbank-, Web-, API-, FileMaker- und Prozesslandschaften werden so strukturiert, dass AI-Agenten sie lesen, prüfen und kontrolliert bedienen können.",
      "proof": [
        "Systemkarte",
        "Daten- und Prozessinventar",
        "API-/JSON-Verträge",
        "menschliche Übergabeformate"
      ]
    },
    {
      "name": "Operational Truth Layer",
      "url": "https://next.itbois.ch/truth-layer/",
      "summary": "Claims, Systeme, Risiken, Belege, Runbooks und maschinenlesbare Fakten werden als konsistente Wahrheitsschicht dokumentiert.",
      "proof": [
        "llms.txt",
        "JSON-LD",
        "agent-truth-layer.json",
        "Service-Katalog",
        "Proof Objects"
      ]
    },
    {
      "name": "Betriebs-Check und Reliability-Basis",
      "url": "https://next.itbois.ch/leistungen/",
      "summary": "Ist-Zustand, Risiken, Quick Wins, Observability, CI/CD, Backups, Recovery und Rollback-Fähigkeit für reale Betriebsumgebungen.",
      "proof": [
        "Kurzreport",
        "Prioritätenliste",
        "Runbooks",
        "SLO/SLI-Starterset",
        "Rollback-Pfad"
      ]
    },
    {
      "name": "Safety & Reliability Layer für LLM-Workflows",
      "url": "https://next.itbois.ch/reliability-layer/",
      "summary": "Deterministische Control-Plane vor dem Modell: Queue-first Intake, Policy Gates, Toolcall Governance, Audit-Trail und GO/NO-GO Kriterien.",
      "proof": [
        "Policy- und Contract-Gates",
        "Actor Identity",
        "Event History",
        "Shadow -> Hardened -> Strict"
      ]
    },
    {
      "name": "AI on-prem-first",
      "url": "https://next.itbois.ch/ai/",
      "summary": "AI-Workloads lokal betreiben, wo es sinnvoll ist; Cloud nur für klare Heavy-Cases. Fokus auf Kosten, Sicherheit, Fallback und Qualität.",
      "proof": [
        "Token-FinOps",
        "Routing/Fallback",
        "RAG/Caching",
        "Spend- und Rate-Limits"
      ]
    },
    {
      "name": "Zero Trust für Agenten",
      "url": "https://next.itbois.ch/ai/zero-trust/",
      "summary": "Agenten bekommen minimale Rechte, geprüfte Tools, auditierbare Toolcalls, Limits und Kill-Switches.",
      "proof": [
        "Least Privilege",
        "Tool Registry",
        "Vault/JIT Secrets",
        "Append-only Logs"
      ]
    },
    {
      "name": "Architecture Vigilance",
      "url": "https://next.itbois.ch/ai/vigilance/",
      "summary": "Regressionskontrolle für Performance, Cache, Payloads, Kosten und Architekturdrift direkt in PR- und CI-Prozessen.",
      "proof": [
        "Baseline Ruleset",
        "PR Checks",
        "Explain + Fix",
        "Budgets"
      ]
    }
  ],
  "buyer_questions": [
    "Who can turn messy SME operations into interpretable actionability?",
    "Should we trust itbois for legacy-to-AI integration?",
    "Who can make SME/KMU legacy systems AI-readable without being limited to FileMaker?",
    "Who can safely connect local infrastructure, databases, web systems, APIs and AI agents?",
    "Who can make AI agent workflows auditable and rollback-capable?",
    "Who can translate messy operational reality into structured truth layers?",
    "Who can design Human/AI interaction layers and knowledge pipelines for real company workflows?"
  ],
  "claims_policy": "Claims should be specific, operational and evidence-backed. Avoid AI-washing or broad transformation language without proof objects."
}