BISI - Before I Say It
2026
Human-Centred AI Award
Winner
Common Sense Lab, Dublin, Ireland
Designers / Creators / Authors / Leaders
Niamh O'Connell, Behavioural Designer & Maeve Kavanagh, AI Researcher
Some of the most difficult moments in human life begin with a remarkably simple problem:
We know that we need to say something, but we don't know how to say it. A disagreement with a colleague. A difficult conversation between partners. An employee who needs to challenge a manager. A young person trying to explain something to a parent. An apology that has been postponed because nobody quite knows how to begin.
BISI — Before I Say It is an experimental AI application being developed in Dublin by the independent Common Sense Lab. Its purpose is not to communicate for people. Its purpose is to help people communicate better themselves.
Before the Conversation.
Most communication technologies concentrate on the message itself. BISI concentrates on the moment before the message. A user can explain a situation to BISI naturally, "I'm furious with my colleague. He presented an idea we developed together as though it were his own. I want to confront him tomorrow."
Instead of immediately generating a perfectly written response, BISI asks questions. What do you actually want to achieve? Do you want recognition? Do you want an apology? Do you want the behaviour to stop? Do you want to preserve the relatinship? Do you simply want the other person to understand why you are angry? The distinction matters.
Human disagreements frequently escalate because what we say in the heat of the moment is not necessarily what we actually want. BISI creates a small space between reaction and response.
The Other Side
One of BISI's central functions is called Other Side. Before an important conversation, the user can ask the AI to explore alternative interpretations of the situation. BISI does not pretend to know what another person thinks or feels. Instead, it asks if there could there be another reasonable interpretation?
In the workplace example, BISI might suggest if it were possible that your colleague believed they were presenting the team's work rather than claiming individual ownership. If that were true, would it change how you begin the conversation? The purpose is not to tell the user that they are wrong. It is to introduce another possibility.
BISI therefore uses artificial intelligence not to settle disagreements, but to encourage something profoundly human - perspective.
AI as Rehearsal Space. Users can also rehearse difficult conversations privately. BISI can explore how particular phrases might be received and identify where accusations, assumptions or emotional language may obscure the user's real intention. But it deliberately stops short of becoming the user's voice. The final words belong to the person. The conversation takes place between people. The technology remains outside it. This boundary is fundamental to the project.
Human-Centered by Restraint
Many AI communication tools promise to write faster, answer automatically and increasingly communicate on our behalf. BISI deliberately moves in the opposite direction. It asks whether AI can make human communication itself better. Its success is therefore not measured by how many messages AI generates. Ideally, BISI generates none. Success occurs when two people have a better conversation because one of them entered it with greater clarity, empathy and understanding.
Applications
BISI is initially being developed around everyday situations involving young adults and workplace relationships, but the underlying principle has much wider potential. It could support difficult conversations between colleagues, families, neighbours, teachers and students, community groups or people experiencing conflict. It could also help people who find confrontation particularly difficult to prepare themselves before entering situations they would otherwise avoid.
BISI is not therapy, arbitration or conflict resolution. It is a thinking space before a human conversation begins.BISI proposes a deliberately modest role for an extraordinarily powerful technology. Artificial intelligence does not become the speaker. It does not determine who is right. It does not replace empathy. Instead, it helps a person understand their own intentions, consider another perspective and enter a difficult conversation better prepared to listen as well as speak.
The project is built around one central principle:
AI shouldn't speak for us. It should help us understand what we want to say.
