Why this document

GenAI can produce convincing text in seconds. Even so, we believe the most important thing in designing products and services is understanding the why behind the choices we make. Flawless content that doesn't respond to people's real needs only makes wrong solutions harder to question.

AI is an incredibly powerful tool. But it's still a tool. And the responsibility for what we design remains ours. 

This document describes how we're using GenAI in our work, to be transparent with our clients, and to contribute, in a small way, to building good practices alongside the professional community we're part of.

How we use GenAI in our work

In minuscola ci occupiamo di Content Design: progettiamo contenuti che mettono in relazione i bisogni delle persone con gli obiettivi di business. Ogni progetto attraversa due momenti: prima divergiamo, per capire il contesto, il business, le persone, poi convergiamo, per trasformare quello che abbiamo imparato in scelte di progettazione concrete. In alcune fasi di questo processo usiamo strumenti di GenAI.

Discovery and research
This is when we need to understand. We use AI for domain research, content audits, industry benchmarking, and analysis of interviews with users and stakeholders.

Synthesis and reporting
This is when we need to condense and organise what we've gathered. We use AI to generate summaries, reports, and presentations from the notes, data, and insights collected throughout a project.

Design and testing
This is when we need to converge and translate insights into design decisions. We use AI to generate wireframes and prototypes to test with real users.

Tools we use
We currently work mainly with Claude (Anthropic) and Perplexity. We also use other platforms with integrated AI features: Condens for interview analysis, Google Stitch for designing interfaces and prototypes. We keep exploring new tools with the same critical approach we bring to everything else.

When we don't use GenAI

To define strategic and business objectives
This is when the right questions matter most: direct conversation with stakeholders where we map the project scope, goals, challenges, and constraints. No tool can replace that conversation.

To conduct usability tests
Tools that delegate user interviews to AI agents already exist, but we believe a researcher's ability to conduct them well isn't replaceable, not yet.

When processing documents or files containing sensitive data
Before uploading any material to GenAI tools, we always verify it contains no personal data or confidential information.

To send reports or documents without review
No AI-generated output is delivered to a client or collaborator without a thorough review from us. To make that review more consistent, we're building a simple internal checklist.

Our principles

We continuously test, evaluate, and reflect on the limits and opportunities these tools offer, with the goal of improving our processes and outcomes, not replacing professional judgement.

Responsibility
The responsibility for what we produce is always ours, regardless of how we produce it. AI's speed doesn't reduce that responsibility, it makes it more deliberate.

Verification
We always check sources and assess the quality, accuracy, and relevance of generated content against the project's objectives.

Consistency
Every AI-generated output is evaluated against the specific goals of the project. Objectives and strategic direction guide how we use these tools, not the other way around.

Inclusivity and accessibility
GenAI tends to reproduce biases present in the data it was trained on. We systematically check that outputs don't contain unnecessary technical jargon, stereotypes or discriminatory language, examples that don't reflect the diversity of people, or accessibility barriers in texts and formats.

Transparency
We let our clients and partners know when GenAI has been used as part of a process.