I built an AI tool because I was tired of the copy-paste dance. Yoink AI writes directly in your apps. Here's what I learned

Like many of you, I've been using ChatGPT/Claude daily. But I kept catching myself thinking: "I can do this faster than AI."

The workflow was killing me: Copy context → Switch tabs → Paste → Edit prompt → Copy response → Switch back → Paste → Edit output

By the time I did all that, I could've just written it myself. So I built Yoink AI - press ⌘⇧Y anywhere on Mac, and AI writes directly where you're working with context of whatever you're already doing.

The Problem That Led Me Here

I tracked my workflow for a week. I was spending ~5 minutes per hour just on the AI dance:

- copying/pasting

- switching between apps

- editing AI's verbose responses

What I Built

Yoink AI captures context from wherever you're working (via accessibility APIs, direct integrations and sometimes screenshots) and writes directly in your app. No tabs, no copying.

Key features:

- Universal hotkey: ⌘⇧Y works in any Mac app

- Smart context: Grabs text fields, documents, even visual elements

- Direct insertion: AI output goes straight to your cursor

- Voices: Custom writing styles. Provide examples of voices you like, and we will save it so you can re-use those voices easily

### Early Results & Learnings

We launched early and have been getting a ton of feedback. Got to 100+ users in <1 week and growing steadily now. A few of paid users too.

What's working:

- Most used in: Email (35%), Docs (31%), Others (34%)

- Surprise use case: People using it to do all sorts of stuff, like Instagram Captions, Twitter Posts etc. One of the feedback sessions had someone using it to proofread powerpoint slides!

What failed:

Our first version didn't have fail safes and assumed that the app would work as intended all the time (e.g. what if the API request to OpenAI was slow this time?) Frustrated a few early users - an app that is supposed to fit your workflow can't only work some of the time!

Technical challenges:

• Accessibility APIs are a nightmare across different apps

• Had to build fallbacks for everything

• Screenshot OCR saved us when APIs failed

Pricing Strategy

Went against the "charge more" advice initially, but still learning a ton here.

Free tier: 10 requests/month and up to 3 voices

• Enough to test value

• Forces us to nail the experience

Pro: $20/month unlimited

• Priced like Grammarly, but broader use case

Questions for the Community

1. Distribution: What channels worked best for your Mac apps?

2. Retention: How do you handle the "I forgot this exists" problem with menu bar apps?

3. Voices feature: We're letting users create writing styles from examples. Too complex or useful differentiator?

What's Next

• Better context management and being able to re-use context seamlessly

• Better integration with Browser-based apps (e.g. Google Docs)

• Deeper app integrations (Notion, Slack, VS Code)

If you want to try it, I'd love feedback from this community. You all know the struggle of building in public.

Ask me anything about building Mac apps, dealing with Apple's accessibility APIs, or pivoting based on user feedback. Happy to share what I've learned!

P.S. - Yes, I used Yoink AI to help write parts of this post. Meta, I know.