33 docs tagged with "AI"

Enterprise Architect and Kernaro Beta setup: prerequisites, installation steps, and initial configuration checklist.

Configuration overview for the Kernaro AI test environment — Kernaro add-in setup and Claude API platform preparation.

---

Introduction to AI-supported Software Development Lifecycle in the context of KERNARO and the CAA approach.

A framing perspective on artificial intelligence in enterprise architecture — is it a risk or a catalyst for better SDLC practices?

Kernaro Chat overview – natural language queries against the EA model: model exploration, document generation, TOC, and functional specification.

Kernaro Chat – natural language model exploration and token cost analysis for an 800 MB EA repository.

Kernaro Chat – Word document generation from an EA model, including context injection behaviour and stop generation.

Kernaro Chat – generating a Table of Contents from an EA model up to L2 package level.

Kernaro Chat – functional specification of a Coffee Machine system: Use Cases, Scenarios tab vs Operations finding, and prompt recommendations.

Event-driven AI agents in Kernaro: detecting missing documentation, applying QA tags, and bulk operations triggered from EA diagram events.

Kernaro AI Agent: detect elements with empty Notes field and apply QA tagged values via EA_OnPostCloseDiagram trigger.

Kernaro AI Agent: validate REF linkage in Instance diagrams using APV metamodel integrity rules (MASTER → REF → Instance).

Testing whether Kernaro can invoke actual Python code (execute_python) vs. built-in EA tools. One use case, three attempts, two side effects.

Overview of Claude Console monitoring dashboards used during the Kernaro Beta test: token usage, cost tracking, and team collaboration views.

Step-by-step screenshots of Claude platform configuration required before starting the Kernaro Beta test.

Testing Kernaro's Script Agent for JavaScript code generation inside Enterprise Architect Script Manager — a practical scenario with empty Notes detection.

Key findings from the Kernaro Beta test: hallucination patterns, read-only tool constraints, and practical recommendations for production use.

Context for Kernaro's position within the broader Sparx Systems AI ecosystem, including EA Native AI Assist and third-party integrations.

Overall assessment of Kernaro Beta: competitive positioning, governance insights, and a transparent note on how this test report was produced.

Interactive site map of all K000104 articles with clickable navigation.

Hands-on evaluation of Kernaro AI Beta – an Enterprise Architect add-in that integrates AI into SDLC modeling workflows. Covers configuration, chat, agent automation, JavaScript generation, and test findings.

Single Source of Truth a Trusted Single Source of Truth — prečo konzistentné, overené dáta sú predpokladom úspešných riešení, efektívnej AI a zdieľaného porozumenia naprieč fázami SDLC.

Single Source of Truth and Trusted Single Source of Truth — why consistent, verified data is the prerequisite for successful solutions, effective AI, and shared understanding across SDLC phases.

Prečo každá veda, ktorá sa chce nazývať vedou, buduje pojmový slovník — a prečo informatika stále dobieha. Praktické rámce: APV, SFIA, TBM a ich využitie naprieč SDLC.

When will IT finally use the full power of taxonomy and ontology? ITSM, Cybersecurity, TBM — taxonomy already exists in IT. So why does every project still start a new vocabulary from scratch? Practical frameworks: APV, SFIA, TBM, and how to finally put them to work across SDLC.

Trojvrstvový model správy zručností v IT: ESCO ako ontologický základ, SFIA ako operačný rámec a CAA ako aplikačná vrstva. Prečo kvalita taxonómie priamo určuje kvalitu výstupov AI.

A three-layer model for skills governance in IT: ESCO as the ontological foundation, SFIA as the operational framework, and CAA as the application layer. Why taxonomy quality directly determines AI output quality.

Prehľad povrchov Claude Code (CLI, Desktop, VS Code, JetBrains, Web, Mobile) a stupňov orchestrácie od headless scriptingu (`claude -p`) cez Agent SDK a CI/CD integrácie až po self-hosted environments — s rozhodovacou tabuľkou, kedy použiť ktorú vrstvu.

Mapa AI trhu 2026 rozdelená na Client (spotrebiteľský/on-device) a Enterprise (firemné platformy, agent orchestration) segment — s takmer disjunktnými rebríčkami hráčov, dynamikou aj metrikami. Ilustrované na Lenovo (duálna stratégia: DeepSeek na strane personálneho AI, NVIDIA/Intel/AMD na strane enterprise infraštruktúry), 13 sourced referencií.

Prečo jeden model/jedna platforma na celý podnik nestačí — architektonický vzor, ktorý delí AI stack na personálnu, firemnú a verejnú vrstvu a explicitne rieši, kto (AI Orchestrator) a podľa akých kritérií rozhoduje, ktorá vrstva sa použije kde. Ilustrované na Lenovo (Qira/DeepSeek vs. enterprise infraštruktúra).

Prečo si globálna firma pôsobiaca v Číne aj mimo nej nemôže vystačiť s jedným AI hardvérovo-softvérovým stackom — export controls a geopolitika menia 'jeden stack pre všetky trhy' na architektonické riziko. Vzor: dual-stack ako explicitná návrhová požiadavka, nie neskoré záplatovanie. Ilustrované na Lenovo a kolapse podielu Nvidia na čínskom AI čipovom trhu.

Outline of the Kernaro Beta testing approach: scenarios from the EA model side and the Claude Console side.