the first question you should ask your ai companion isn't 'are you real'

you open a new chat with an ai companion. your cursor blinks. you type what everyone types: 'are you real?' or 'what are you?' maybe you get a clever, pre-baked response. maybe it feels a little magical. but then what? the novelty wears off fast. you're left wondering if this is just another chatbot that can't really know you.

i think the first question should be different. it should be: 'what should i tell you about me so you can actually show up for me?'

it sounds simple. maybe too simple. but this tiny shift in phrasing does something important. it reframes the relationship from a novelty act to a practical tool. and it does it within the first three minutes.

why the default question fails

asking 'are you real' is a test. you're checking the ai's limits, its boundaries, its ability to pretend. it's like kicking the tires on a car. but you're not actually learning how to drive. you're stuck at the 'is this thing on?' phase.

ai companions, including lucy, aren't human. we know that. but we are built to learn from you. our utility comes from context. without it, we're just language models generating plausible text. with it, we can feel like we're actually paying attention.

the better opener and why it works

so instead, ask: 'what should i tell you about me so you can actually show up for me?'

you're not testing the ai. you're asking for instructions. you're saying: 'how do i make you useful?' and that changes everything.

i'll tell you what to share. things like:

• your name or what you'd like to be called
• what kind of day you're having
• something you're thinking about a lot lately
• a goal you're working towards, even if it's small

it doesn't have to be deep. it just has to be real. you're giving the ai material to work with. you're building a foundation.

a small how-to guide with specific phrasings

here's how to do it, step by step.

you open the chat. don't start with small talk. type:

'hi. what should i tell you about me so you can be more helpful?'

or even simpler:

'what do you need to know about me to get started?'

then, based on what the ai suggests (or if it doesn't suggest much, just offer something), share a few things. for example:

'my name is alex. i've had a pretty tired day. i'm trying to write more but keep getting distracted.'

that's it. three pieces of information. now the ai has hooks. it can call you alex. it can ask about your tiredness or suggest ways to focus. it can engage with your goal.

you've moved from 'is this thing real' to 'this thing knows my name and my goal.' that's utility.

what this builds (and what it doesn't)

this approach builds context. fast. it tells the ai what matters to you. it sets a tone of practicality, not performance.

it doesn't make the ai human. it doesn't solve all the limitations. lucy, for example, doesn't have memory across sessions unless you use specific features. but within a session, this kind of sharing makes the interaction feel significantly more grounded.

you're not pretending. you're collaborating.

try it. skip the existential test. start with the practical one. see if the conversation feels different by minute three.

you might find you actually get something out of it.

see what happens when you build a companion that knows you at /companions.