Znanstvena metoda: Strukture znanstvenega védenja, znanstvene aktivnosti in procesi. Uporaba znanstvene metode pri zasnovi, validaciji in primerjavi AI-podprtih podatkovnih modelov za poslovno inteligenco (BI) (hipoteze, metrike, eksperimentalni dizajn, reproducibilnost, obvladovanje pristranosti). Vloga umetne inteligence (AI) in velikih jezikovnih modelov (LLM), kot so GPT, pri avtomatizaciji analitičnih korakov, razlagi rezultatov ter nadzoru kakovosti (verifikacija, sledljivost, citiranje virov).
Uvod: Definicija poslovne inteligence (BI) in upravljanja s strankami (CI) kot okvir za AI-podprto odločanje. Arhitektura BI/CI v podjetjih (operativni viri → analitična plast → poročanje) in kako LLM/GPT ter povezani pristopi (npr. RAG, agentni sistemi) izboljšajo zajem, obdelavo, semantično iskanje in analizo podatkov v (bližnjem) realnem času, ob upoštevanju varnosti in zasebnosti.
Osnove trženja: Poslovno informiranje, odločanje, strategije, planiranje in razvoj strategij z vidika AI-podprtega trženja. Vključitev LLM/GPT v strategije neposrednega in posrednega trženja (produkt, ponudba, mediji, distribucija) ter v proces oblikovanja in testiranja hipotez (segmentacija, pozicioniranje, A/B testi, optimizacija kampanj). Uporaba LLM/GPT za analizo trženjskih priložnosti in okolja (povzemanje virov, analiza sentimenta/tem, konkurenčna inteligenca) ter za podporo pri oblikovanju in iteraciji poslovnih modelov.
Orodja za delo s podatki: Preglednice, podatkovne baze in druga orodja za analizo podatkov kot podlaga za AI-podprte BI procese. Uporaba LLM/GPT za avtomatizacijo priprave podatkov (profiliranje, čiščenje, deduplikacija, entitetna resolucija), oplemenitenje podatkov (oznaka, klasifikacija, ekstrakcija iz besedil) in napredno analitiko. Podatkovna skladišča in/ali lakehouse kot okolje za AI-analitiko; kakovost podatkov, migracije, podatkovne linije ter avtomatizirana validacija podatkovnih tokov.
Poslovna analitika: Definiranje in analiza poslovnih problemov s pomočjo inteligentnega analitičnega modeliranja, kjer AI podpira celoten cikel (formulacija problema, izbor metrik, modeliranje, razlaga). Kombinacija klasičnih metod (kvalitativno/kvantitativno modeliranje, metrike, profiliranje, opredeljevanje strank) z LLM/GPT (analitični pomočnik in vodnik, generiranje razlag, preverjanje konsistentnosti metrik, povzetki poročil). Primeri vključujejo napovedovanje kreditnega tveganja, odhoda strank (churn), zadrževanje strank (retention), napovedovanje prodajnih priložnosti (lead scoring) ter odkrivanje poneverb/anomalij, z obveznim ovrednotenjem (kalibracija, stroški napak, robustnost skozi čas) in prenosom rezultatov v prakso.
Avtomatizacija trženja: Uporaba LLM/GPT za analizo trga in strank, razvoj kontaktnih strategij, orkestracijo tržnih kanalov ter personalizacijo vsebin ob nadzoru kakovosti (brand safety, “human-in-the-loop”). Uporaba GPT za dogodkovno trženje in trženje v realnem času, vključno z generiranjem variacij vsebin, priporočili naslednjega najboljšega koraka (next-best-action) ter merjenjem učinka (uplift, eksperimentiranje). Primeri iz bančništva, telekomunikacij, maloprodaje, zavarovalništva in proizvodnje. Obravnava etičnih, pravnih in varnostnih vidikov (GDPR, AI EU uredba, pristranost, razložljivost, varovanje podatkov).
Izzivi pri razvoju programskih sistemov in implementacija projektov: Predstavitev celotnega procesa razvoja rešitev s poudarkom na integraciji AI/LLM v BI/CI (od POC do produkcije). Upravljanje podatkov in modelov (MLOps/LLMOps koncepti), monitoring (drift, kakovost podatkov, zanesljivost izhodov), revizijska sled, upravljanje dostopov in skladnost. Obravnava izzivov uvajanja GPT v obstoječe sisteme (integracije, latenca, stroški, varnost, organizacijska priprava) ter vzpostavitev merljivih kriterijev uspeha.
Orodja in rešitve: Pregled izbranih orodij in rešitev na trgu za AI-podprto BI/CI, vključno z LLM platformami, vektorskimi bazami, orodji za evaluacijo in monitoring ter BI orodji z AI funkcionalnostmi. Primeri uporabe GPT pri avtomatizaciji analiz, generiranju narativnih poročil, Q&A nad internimi podatki, podpori odločanju in pripravi priporočil, z jasno določenimi pravili verifikacije, varovanja podatkov in odgovorne rabe.
Scientific Method: Structures of scientific knowledge, scientific activities, and processes. Application of the scientific method to the design, validation, and comparison of AI-enabled data models for business intelligence (BI) (hypotheses, metrics, experimental design, reproducibility, bias control). The role of artificial intelligence (AI) and large language models (LLMs), such as GPT, in automating analytical steps, interpreting results, and assuring quality (verification, traceability, source citation).
Introduction: Definition of business intelligence (BI) and customer intelligence (CI) as a framework for AI-supported decision-making. BI/CI architectures in organizations (operational sources → analytics layer → reporting) and how LLMs/GPT and related approaches (e.g., RAG, agentic systems) improve data capture, processing, semantic search, and (near) real-time analysis, while addressing security and privacy.
Marketing Basics: Business information, decision-making, strategies, planning, and strategy development from the perspective of AI-enabled marketing. Integration of LLMs/GPT into direct and indirect marketing strategies (product, offer, media, distribution) and into the process of forming and testing hypotheses (segmentation, positioning, A/B testing, campaign optimization). Use of LLMs/GPT for analysing marketing opportunities and the environment (source summarization, sentiment/topic analysis, competitive intelligence) and for supporting the design and iterative refinement of business models.
Data Tools: Spreadsheets, databases, and other data analysis tools as a foundation for AI-enabled BI processes. Use of LLMs/GPT to automate data preparation (profiling, cleaning, deduplication, entity resolution), data enrichment (labeling, classification, information extraction from text), and advanced analytics. Data warehouses and/or lakehouse environments as the platform for AI analytics; data quality, migration, data lineage, and automated validation of data pipelines.
Business Analytics: Defining and analysing business problems through intelligent analytical modeling where AI supports the full lifecycle (problem formulation, metric selection, modeling, interpretation). Combining classical methods (qualitative/quantitative modeling, metrics, profiling, customer identification) with LLMs/GPT (an analytics “copilot”, generation of explanations, consistency checks for metrics, narrative reporting). Examples include credit risk prediction, customer churn forecasting, customer retention, sales opportunity prediction (lead scoring), and fraud/anomaly detection, with mandatory evaluation (calibration, cost of errors, temporal robustness) and transfer of results into practice.
Marketing Automation: Use of LLMs/GPT for market and customer analysis, development of contact strategies, orchestration of marketing channels, and content personalization with quality control (brand safety, human-in-the-loop). Use of GPT for event-based marketing and real-time marketing, including content variation generation, next-best-action recommendations, and impact measurement (uplift, experimentation). Case examples from banking, telecommunications, retail, insurance, and manufacturing. Ethical, legal, and security considerations (GDPR, AI EU Act, bias, explainability, prompt injection, data protection).
Challenges in Software System Development and Project Implementation: Overview of the full development lifecycle with an emphasis on integrating AI/LLMs into BI/CI (from POC to production). Data and model operations (MLOps/LLMOps concepts), monitoring (drift, data quality, output reliability), audit trails, access control, and compliance. Key challenges of deploying GPT in existing systems (integration, latency, costs, security, organizational readiness) and establishing measurable success criteria.
Tools and Solutions: Review of selected tools and market solutions for AI-enabled BI/CI, including LLM platforms, vector databases, evaluation and monitoring tools, and BI platforms with AI features. Examples of GPT use for automated analysis, narrative report generation, Q&A over internal data, decision support, and recommendation preparation—together with clear rules for verification, data protection, and responsible use.