AI in working with SQL – practical use of LLM models and agents for data analysis

Training process

Training needs analysis

If you have specific requirements regarding the training programme, we will carry out a training needs analysis for you. This will guide us on which aspects of the programme should receive greater emphasis, so that the training programme meets your specific needs.

What will you gain?

Smarter LLM use in analytics - You will learn when LLMs truly speed up data analysis and when they introduce unnecessary risk, so you can use AI deliberately, with better judgment and more reliable outcomes.

Faster SQL work - You will turn business questions into solid SQL faster, refactor existing queries more confidently, and catch edge cases that often distort analytical results or reporting output.

Effective analytics prompts - You will master prompts for data tasks, choose the right input format, and set clear model constraints so the responses you get are more precise, useful, and easier to validate.

Fewer wrong conclusions - You will learn how to spot hallucinations, anomalies, and inconsistent numeric outputs, making it easier to stop flawed insights before they reach a dashboard, deck, or business decision.

Stronger insight delivery - You will use AI to summarize analyses, generate product hypotheses, validate KPIs, and interpret results, so you can prepare outputs that are clearer and more useful for business partners.

Support for A/B testing - You will learn how to use AI in experiment design, hypothesis framing, and statistical interpretation without blindly accepting automated conclusions produced from test results.

AI in your daily workflow - You will see how to connect language models with databases, code, RAG, and day-to-day analyst tasks, helping you shorten ad hoc analysis, documentation, and stakeholder communication.

Safer work with data - You will learn Local-First Analytics, anonymization methods, and hybrid approaches, so you can protect sensitive data better and choose the right model setup for your organization.

Training programme

1. Introduction to AI and LLM models in the analytical context

what LLM models are and how they work,
deterministic vs probabilistic AI,
strengths and limitations of LLMs in data analysis,
where AI really helps the analyst and where it harms,
the evolution of the analyst's role: from „query operator” to „systems orchestrator”.

2. Token economics, context management and model selection for analytical tasks

token economics,
context window management,
RLM models,
model selection for the task,
AI systems architecture,
connecting the language model with the database and code.

3. Prompt Engineering for data analysis – fundamentals of effective communication with LLM models

how an LLM model „thinks” and what this means for prompts,
structure of an analytical prompt,
context engineering,
preparation of data in Markdown, JSON, XML and CSV,
system prompt,
defining roles, constraints and output formats.

4. Prompt Engineering in work with SQL and business requirements

generating SQL queries,
refactoring SQL queries,
translating business requirements into analytical queries,
analysis of existing queries,
readability and optimization,
identification of edge cases.

5. “Memory” of AI systems and the use of additional knowledge sources in analyses

RAG,
Agentic RAG,
vector databases,
graph databases,
providing context to the model,
use of organizational data in AI systems.

6. Analytical cases – working with data, insights and KPIs

automatic summaries of analysis results,
generation of product insights,
generation of hypotheses,
sanity checks,
detection of anomalies, gaps and illogical results,
creation and validation of KPIs,
identification of potential errors in data and analyses.

7. A/B tests and experiments using AI

AI support in designing experiments,
formulating test hypotheses,
interpretation of statistical results,
supporting inference and communication of results,
risks of automatic interpretations of A/B tests,
„synthetic users” and AI Personas.

8. Workshops on the team's real scenarios and integration of AI with the analyst's daily work

working on real examples from the team,
optimization of existing analytical prompts,
testing different variants of prompts,
comparing results and the quality of responses,
assessment of the usefulness of AI in specific use cases,
AI as the analyst's „copilot”, not a substitute.

9. Using AI in the daily analytical workflow and automation of repetitive tasks

ad-hoc analyses,
data exploration,
analysis documentation,
communication of insights to the business,
automation of repetitive analytical tasks,
best practices of teamwork with AI.

10. Quality, ethics, security and the recommended way of working with AI in an analytical team

validation of results generated by AI,
the analyst's responsibility for the results,
methods of detecting hallucinations in numbers and facts,
auditability and Explainable AI,
LLM-as-a-Judge,
open-source models: Llama 3, Mistral, Gemma,
Ollama and LM Studio,
Local-First Analytics,
data anonymization and masking,
hybrid approach,
good organizational practices,
when to use AI and when to use classical methods,
checklists for analytical work with AI,
recommended workflow of an analyst supported by AI,
areas for the team's further development.

What are the prerequisites for participating in the training?

SQL basics - You should be comfortable reading and writing simple SQL queries, including SELECT, JOIN, WHERE, GROUP BY, and aggregations, to fully benefit from the hands-on exercises.

Analytics experience - You should have practical experience in data analysis or reporting so you can follow business questions, KPIs, and common data quality issues discussed during the training.

Working with data - You should understand tables, columns, filters, and common data formats such as CSV or JSON, so you can work smoothly with the examples and prompts used in the course.

Basic test statistics - You should know core concepts used in experiment analysis, such as hypothesis, metric, and significance, so you can properly follow the section devoted to A/B testing.