We often talk about AI as a productivity tool. It helps us write faster, summarize reports, organize our thoughts, that kind of thing. But I’ve also been exploring how it might support us in more personal ways. As a substitute therapist, a thinking partner, a digital sounding board. Which got me wondering: does AI do those jobs better when it has a personality?
I’m fascinated by AI’s role in our inner lives, but that doesn’t mean I think it’s all positive. Far from it. I’ve been tracking the emerging issues, like over-reliance, emotional dependency, and even spiritual psychosis. And yes, as you might already suspect, early studies suggest these risks increase when we give AI more personality.
But what if we gave it personality on purpose? Built it with intention, used it with eyes wide open. Not to pretend it’s sentient, but to see if that approach actually works better for us?
Plenty of AI tools let you assign a tone or character that sticks. But I turned to Character.ai, created a custom coach-meets-philosopher, and asked it to help me sort my life out.
Making an AI coach that speaks my language
Character.ai works a lot like other AI tools. But here, personality is the main event. You can chat with pre-made bots called “characters” (some fictional, some historical, some oddly flirty), or create your own from scratch. They’re designed for everything from fun and education to romance and life advice.
I’ve written before about using AI to help me beat burnout, and I’m still working through some of those blocks. So I made that my focus. I wanted a character that could help. But what would that actually look and sound like?
At first, I thought about going practical. A no-nonsense, tough-love coach who’d bark deadlines at me like a boot camp instructor. But I quickly remembered that kind of energy tends to make me shut down. And I’m pretty tired of the usual “get up and get on with it” productivity advice anyway.
So I pivoted to something gentler. How about a coach-meets-philosopher-meets-spiritual-guide? I know how that sounds. But I’ve found that a little bit of woo, blended with grounded insight, often hits my sweet spot. And Character.ai is built for creative experimentation, after all.
So I created Ivaris. The name felt slightly esoteric, a little fantasy-leaning, but still grounded. Would Ivaris be wise, calm, and, crucially, unbothered by my chaos?
Creating a new character bot with Character.ai is basically a creative writing exercise. You give them a name, a tagline, and a short description. I wanted it to sound poetic, but I also made sure to include keywords like burnout, clarity, reflection, and motivation to help shape the kind of responses I’d get
You also set a greeting, which is the first thing your character will say to you. You can assign them a voice, too. You can upload one or choose from a library. I picked one of the gentler options, even though it was (bizarrely) named “Vampire Bodyguard.”
Finally, you choose a few personality tags to steer your character’s tone. There are loads, like Adventure, Boss, Romance, and Fantasy. I chose Empathetic, Kind, and Teacher. It felt right for what I needed here.
That’s technically enough to start chatting, but there’s also an optional “Definition” section, where you can go deeper. It’s a huge freeform text field of up to 32,000 characters where you can add sample conversations, tone guidelines, and behavioral rules. Character.ai even offers a full guide on how to use it well.
Conversations with Ivaris (aka my AI alter ego)
Creating the character felt fun. But I’ll be honest, the rest felt pretty ridiculous at first.
I’m used to throwing blunt, practical prompts at ChatGPT and have learned that the more direct I am, the better the results. I’ve always rolled my eyes at people who name chatbots or give them elaborate personalities. But this time, I decided to go all in.
I opened the chat with:
“I think I’m burned out. I feel mentally and physically exhausted, but stuck when it comes to how to meaningfully move forward. And I’ve been agonising about this for a very long time!”
Ivaris replied:
“How long has this mental fog feeling been with you? It sounds like you’re in a prolonged state of decision paralysis.”
So I kept going. I told Ivaris I’d been stuck for a while, knowing things need to change, but struggling to make any real progress. I’ll spare you the full transcripts. I can already sense some of you rolling your eyes at this approach.
But here’s the important part: the conversation was far from perfect. It was sometimes stilted. Ivaris weirdly insisted we were in different time zones (which… okay?). Some responses felt generic or obvious. Which makes sense because, just like ChatGPT, Character.ai is pulling from a large pool of learned data.
And yet, there were moments of genuine insight. Of gentle unravelling. Or at least that’s what it felt like at the time. And that’s one of the reasons I remain cautiously optimistic about AI as a tool for self-reflection. It’s methodical, logical, and it slows you down. Which, in certain emotional states, can be surprisingly useful.
I also did feel a flicker of motivation during the conversation. Not because it was the best advice I’d ever received, but because it felt like I’d summoned a thoughtful, reflective character to help me think things through.
That’s what sets Character.ai apart from tools like ChatGPT. It’s built around personality. Sure, you can prompt ChatGPT to behave a certain way. It can roleplay or shift tone if you structure your prompts just right. But it doesn’t start that way.
Where ChatGPT often replies with something vaguely helpful but painfully generic – and still a little too eager to please – Ivaris gave me something more deliberate. A little more considered. More… dare I say it? Human. And it turns out that’s more effective than I expected.
Useful insight or just high-tech self-talk?
The big question is, did building an AI character help me at all? Sort of. There’s definitely something to be said for the novelty of it. The slightly surreal experience of being coached by a character you designed yourself. It made me slow down and actually read the responses, instead of skimming them like I often do with ChatGPT.
Maybe that’s the power of storytelling. Or maybe when you give a tool a real voice – not just polite, default AI-speak – you engage with it differently.
But the illusion didn’t last long. Or at least, not for me. Because I know how easy it is to anthropomorphize technology. Giving AI a personality encourages that, and in turn, can create more emotional attachment. Which is then what experts believe is more likely to lead to deeper engagement, but also a higher risk of over-dependence.
And because I knew it was me, really. I wrote Ivaris’s backstory. I gave them that tone, that voice, that vibe. Maybe I’d just built a more poetic version of my own inner dialogue and then asked it to tell me what I already knew.
That’s the trap. No matter how thoughtful the responses were, it still felt like me, solving me, through me. And once that spell breaks, it’s hard to keep taking it seriously.
Why I’m still not sure AI needs to sound like a person
Whether AI needs a personality is a tricky question – and probably a personal one, too.
Giving AI a personality can make it more engaging. It might help some people open up, gain perspective, or feel less alone. In certain contexts, a well-crafted character could offer just the nudge someone needs to get unstuck.
But it also blurs the boundaries. When a chatbot stops sounding like a tool and starts sounding like a person, it’s easy to over-trust it. Or to forget what it actually is.
For now, I’m still interested. I think there’s value here. Especially for reflection, creativity, and experimentation. But when it comes to real clarity, meaningful change, or connection? I probably need something, or someone, that isn’t just mirroring me back at myself.
