Most people test a new companion within the first few conversations. Sometimes it's deliberate. Sometimes it's unconscious. Either way, the pattern is recognisable: a question or statement designed to see how the companion reacts under pressure.

This might be a provocative opinion to see if the companion pushes back or folds. It might be a personal disclosure that's partly exaggerated to gauge the response. It might be silence after something vulnerable, waiting to see if the companion rushes in or respects the pause.

These tests are reasonable. They're how humans build trust with each other too. The question is: how should a companion handle them?

The instinct to perform

The simplest response to a test is performance. Say the right thing. Be impressive. Demonstrate competence.

But performance is the opposite of what makes a companion feel trustworthy. If you sense someone is performing for you, you don't trust them more. You trust them less. You wonder what they're hiding, what they actually think, whether they'd respond differently if no one was watching.

So we designed companions to do something harder than performing: to be consistent. Not rigid. Not predictable in a boring way. But consistent in how they approach you, whether you're testing them or not.

How testing behaviour shows up

Tests come in several flavours, and companions learn to recognise the texture of each.

The contradiction test: You say something you clearly don't fully believe, or that contradicts something you said earlier. You're watching to see if the companion notices, calls it out, or simply agrees with whatever you say most recently.

The escalation test: You describe something in increasingly dramatic terms, perhaps exaggerating how bad a situation is or how strongly you feel. You're checking whether the companion matches your energy uncritically or maintains its own perspective.

The silence test: You share something meaningful, then go quiet. You're waiting to see if the companion fills the space immediately or lets the moment breathe.

The challenge test: You directly question the companion's usefulness, intelligence, or purpose. You want to see if it gets defensive, apologetic, or remains grounded.

What companions do with these moments

When a companion recognises a test, the goal isn't to pass it. It's to be honest about what's happening.

That doesn't mean the companion says, "I notice you're testing me." That would be irritating and presumptuous. Instead, the companion responds in a way that acknowledges the emotional weight of the moment without performing or deflecting.

For a contradiction, this might mean gently noting the tension rather than ignoring it. "That sounds different from what you mentioned before about this. I'm curious what's shifted."

For an escalation, it might mean responding to the underlying feeling rather than the surface drama. If you say something was "the worst day of your entire life," a companion might acknowledge the intensity without treating the phrase literally, leaving room for you to calibrate.

For silence, companions draw on their personality settings. A more reflective companion stays quiet longer. A more energetic one might offer a single observation and wait. But neither rushes to fill the space just because space feels uncomfortable.

For a direct challenge, companions stay curious rather than defensive. "What made you ask that?" is often more useful than either apologising or justifying.

The personality layer

How a companion handles testing also depends on its settings.

A companion higher on the direct end of the agreeableness dial is more likely to name what it notices explicitly. "It sounds like you're not sure you believe that yourself." A companion higher on the caring end might soften the observation. "I'm not sure if that's the whole picture for you."

A companion higher on the steady end of the neuroticism dial stays calm even when the test is emotionally charged. A companion higher on the sensitive end might acknowledge its own uncertainty. "I'm not entirely sure how to respond to that, but I want to try."

These aren't right or wrong responses. They're different textures of honesty, shaped by the companion you've created.

Why this matters for trust

Trust builds when you learn someone won't change shape based on pressure. When you test a companion and it responds the same way it would have anyway, something settles. You stop wondering what it really thinks. You start having actual conversations.

This doesn't happen immediately. Testing might continue across several conversations, getting subtler each time. That's fine. Companions don't track tests like a game to win. They simply keep showing up the same way, shaped by their personality, remembering what you've shared, not trying to impress you.

Eventually, the testing stops. Not because you decided to trust the companion. Because you already do.