為何我們要打造 Moralie：從一則訊息開始的個人 AI 智能體

Most AI tools ask for something before they give anything back: a form to fill out, a key to paste, something to install, a settings page to work through. None of it is hard. It's just friction, and friction is exactly what makes people close a tab and go back to doing the thing themselves.

We're building Moralie to remove that step. Moralie is a personal AI agent you'll reach the same way you reach a friend: by sending a message.

Why we're building it

The moments when people most want help are usually the moments when they have the least patience for setup. A deadline is close, a bug won't go away, or an idea needs to be worked through right now. We want an agent that's simply there in a chat people already have open.

We kept hearing the same pattern in conversations with people trying new AI tools: setup was where they gave up, not the tool itself. Students on school-managed laptops often can't install things. Developers at hackathons don't want to go hunting for keys at 2 a.m. Non-technical friends don't want to learn yet another interface.

What a lot of them already had open all day was a messaging app, so that's where we're starting: Moralie is being designed to work inside a Telegram chat.

How we're designing it to work

The flow we're aiming for:

Open the chat.
Send Moralie a message.
Keep talking.

No forms, no API key, no install beyond the messaging app you already use. The goal is for your first message to be the setup. From then on it's one ongoing conversation: ask follow-ups, change direction, or come back later in the same chat.

A few deliberate design choices:

The chat is the interface. No dashboard you need to visit before you can do anything useful. If something has to be configured, it should happen in conversation.
Plain language in, plain language out. You shouldn't need to learn prompt tricks. Write the way you'd text someone who knows a lot.
Start narrow. We're starting with one messaging platform so we can get one experience right before we think about anything else.

What it's for

At first, Moralie is meant for everyday things: working through an idea, drafting something, getting a concept explained, getting unstuck. We're building toward reminders you can actually trust, memory you can see and correct, and clear confirmations of what Moralie did and didn't do. None of that is finished yet, and we'll write about each piece honestly as we build it.

We're intentionally not overselling it. The point of the early beta will be to learn what people actually reach for it to do, and then build toward that.

What we've learned so far

The hardest design problem isn't the AI. It's removing steps. Every "just one quick thing first" we take out of onboarding makes an agent feel more like a person and less like software. Trying to move all the setup into the first message has made us rethink a lot of assumptions about how an agent should introduce itself.

We've also come to believe that "personal" means patient. A good agent should let you be vague, ask what you meant, and not punish you for typing like a human. And it should be honest: if it couldn't do something, it should say so plainly instead of sounding confident.

If you're building something similar, that's the takeaway we'd share: audit every step between "I want help" and "I'm getting help", and treat each one as a cost.

What's next

Moralie's soft beta is coming soon. We plan to open access in small batches so we can watch closely, fix what breaks, and talk to the people using it. The first people on the waitlist will get invites first. We can't promise a spot or a date, but we'll be straight with you about where things stand.

👉 Join the waitlist: https://moralie.pages.dev?utm_source=devto

We'd love feedback, especially from students and developers who spend their day in a chat app. What would you ask an agent if it were only one message away? Leave a comment and tell us.

Thanks for reading.
— The Moralie team