How we use AI (and what we never hand over to it)
AI gives us speed; the judgment is ours. Which tasks AI handles in our projects and which ones always stay in human hands.
- Published
- November 13, 2026
- Reading time
- 2 minutes
- Written by
- The Roshal Lab team

AI is a tool, not the product
We use artificial intelligence like any other work tool, and we don't hide it. It makes us faster at the repetitive parts. But the value of a project isn't in the tool: it's in understanding the problem well and deciding what's worth building.
What we use it for
- Handling volume: reading documents, organizing data, summarizing interviews.
- Spotting patterns and anomalies that would take weeks by hand.
- Drafting copy, documentation and code.
- Prototyping quickly so ideas can be tested before anyone invests in them.
- Automating repetitive tasks that eat up your team's hours today.
What we never hand over to it
- Deciding what gets built and what doesn't.
- Talking to your customers on your behalf without human review.
- Approving a deliverable: everything AI produces is reviewed by a person before it reaches you.
- Responsibility for the outcome.
Transparency in practice
Being transparent isn't about stamping "made with AI" on things. It's making sure that, at every stage, you know which part a tool did and which part a person did.
- We tell you which tools we use and what for.
- We agree with you on which data can be used, and where.
- We explain the limitations of every AI-based solution, not just its benefits.

