LinkedIn / AI-assisted BI

AI is useful when the workflow is still understandable.

I use Claude Code, LLM APIs and connected tools to move from a business question to a working system, with validation and a human decision point where it matters.

Start with the job.

AI enters because a real task is slow, repetitive or difficult to document, not because the tool is fashionable.

Keep the model visible.

Inputs, transformations, checks and outputs should remain legible to the person responsible for the decision.

Leave a review gate.

Automation can prepare, compare and explain. The final business action still needs an accountable human.

A practical workflow

From first question to daily tool.

QuestionWhat is taking too long or remaining unclear?

ContextGather the files, definitions, constraints and people involved.

BuildUse Claude Code, Python, APIs or a small interface to create the first useful version.

CheckTest the output, document the assumptions and keep a human in the loop.

RunTurn the validated workflow into something that works again tomorrow.

Selected systems

Small tools, real friction.

Personal and professional experiments that show how I approach applied AI beyond a chatbot window.

DocumentationPBIP documentation

AI-assisted documentation for Power BI projects and semantic models.

TypeScript ↗Development systemAI development workflow

A structured working environment for repeatable development, data and QA workflows.

Claude Code ↗Case directionAI-assisted BI

How I combine automation, context and review instead of treating AI as an unexplained black box.

Read the method ↗

Continue

See the complete professional overview.

Back to LinkedIn