AI in Consulting: Where Language Models Really Help (and Where They Don't)

AI is not magical – but it is useful

In consulting projects, I deploy LLMs strategically where they deliver measurable value: analyzing large volumes of documents, generating tickets or requirements, suggesting refactorings, and as a sparring partner for architecture decisions.

Where AI shines

• Knowledge aggregation across Confluence, PDFs, code repos.
• Requirements clarification: finding gaps in user stories.
• Code reviews: consistent style checks.

Where AI fails

• When domain knowledge is missing and no one provides it.
• When compliance requirements don't make it into the prompt.
• When output is adopted without review.

My approach

I integrate AI into existing processes instead of building new tools around it. This saves money – and delivers results the team actually uses.