The AI Honesty Problem

How Do We Keep the Writer’s Hands on the Wheel?

Key Takeaway

The greatest risk of AI writing tools isn’t bad output. It’s good-enough output, accepted again and again in small moments, slowly replacing the writer’s own voice and thinking without anyone quite noticing. The antidote is physical placement — mechanical, narrow, precise.

The Opposite Question

Most companies building AI writing tools are asking this question: "How do we make the AI do more so the writer can do less?"

When we built Wretify, we asked the opposite. "How do we keep the writer’s hands on the wheel?"

We found the answer by accident.

The Accident

I was working on an article about suffering. The unavoidable kind. The kind that interrupts the autopilot and insists on being noticed.

I was deep in a conversation with the AI in my sidebar, and instead of typing out what I wanted it to look at, I did something I’d never done before.

I highlighted a sentence from my article, dragged it across the screen, and dropped it directly into the sidebar.

It worked.

I sat back for a second, not to understand the technology. To understand what had just happened to me.

Something felt different. Something I had never felt using Notion AI, because their architecture prevents it by default.

Now, I knew exactly what the AI was looking at. Because I had put it there, with my own hands.

The Uncertainty Nobody Talks About

If you’ve spent serious time working with AI in a writing context, really serious time, hours a day, you know a feeling that nobody talks about directly.

It isn’t frustration. It’s a low-grade uncertainty that sits next to you in every writing session.

You type your question into the sidebar. The AI has access to your whole page. And somewhere in the back of your mind, a quiet voice asks: what is it actually focusing on right now?

That uncertainty lives in you the whole time the AI is responding. It’s small. But it compounds.

Most of us try to solve it the same way. Longer prompts. More specific instructions. "Look at the third paragraph. Focus on the section where I describe…" You’ve typed elaborate explanations of content that already exists, in written form, two inches to your left.

I spent months doing that. Explaining my own writing back to an AI that already had it on the page. That’s not a writing experience. That’s an administrative one.

Precision Without Effort

When I started dragging instead of describing, something shifted that I can only call precision without effort.

When you drag a sentence to the sidebar, you have already done editorial work, before the AI responds to a single word. You read the sentence. You evaluated it. You decided it was the thing worth thinking about right now. Your hands moved it deliberately.

Three acts of judgment, expressed as one gesture. The AI didn’t lead your thinking. It followed it.

And because the scope was one sentence instead of an entire article, something strange happened: I went deeper. This is counterintuitive, but it’s real.

When the AI has access to your whole page, your mind is in a management posture — broad, scanning, slightly tense, trying to anticipate what it might touch. When you’ve handed it exactly one thing, your mind settles into that one thing. You’re not managing anymore. You’re thinking.

Narrow scope doesn’t limit the conversation. It concentrates it.

Seeing less, you see more. And more deeply.

Something About the Hands

There’s something else I noticed and later researched. Something about the hands themselves.

When you physically move something — drag it, place it — the body is involved in a way that pressing a button isn’t. The gesture is deliberate. Proprioceptive. The writer’s hands didn’t step back and hand the wheel to the machine. They initiated. They pointed. They placed.

That physical act of initiation keeps you present in a way that "type your question here" quietly doesn’t.

I know that sounds small. It isn’t.

The Real Risk

The greatest risk of AI writing tools isn’t bad output.

It’s good-enough output, accepted again and again in small moments, slowly replacing the writer’s own voice and thinking without anyone quite noticing. Not a sudden takeover — a slow stray.

The drag gesture creates a structural resistance to that drift, because every single interaction begins with an act only you can perform. The AI cannot initiate. It can only respond to what your hands brought to it.

Mechanical. Narrow. Precise.

Not because it’s limited. Because those three things together are what keep the thinking yours.

Both Directions

In Wretify you can drag content from your manuscript into the sidebar — instead of explaining what you want the AI to focus on, you show it. Precisely. With your own hands.

You can drag content from the AI’s response back into your manuscript — reaching in, finding the one sentence that matters in three paragraphs of good ideas, and placing it exactly where it belongs in your work.

Not the whole response. Not wherever the AI decides. The specific insight, in the exact spot you choose, placed by you.

The writer as curator. The AI as a brainstorming partner, research assistant, and collaborator — not a ghostwriter. Your hands never leaving the wheel.

An Invitation

If you want to feel what that feels like — that specific kind of calm that comes from staying in control without fighting for it — come write with us.

← PreviousKeyboard Shortcuts as Habit FormationNext →Local-First AI: The Ecological Position