4
7 Comments

I got tired of copy-pasting into ChatGPT, so I put the prompt next to the text

Every time I read something worth understanding, the loop was the same: copy the paragraph, open a new tab, paste, retype the same instructions I typed yesterday, read the answer, switch back. Dozens of times a day.

The tab switch is cheap. The context you drop on the way is not. InstaPrompt is my fix. Highlight text on any page and a small button pops up right where you selected.

Click a prompt (or hit Alt+Shift+X) and the answer appears in place. You never leave the page. Three things I decided early:

1. No API key on the first screen. Built-in AI, ~25 free uses a month, working the second you install. Every tool I closed the tab on asked for an OpenAI key before showing me anything.

2. Context Labels. Save your tone, brand voice, or research focus once. It attaches to any prompt and travels with every request. That was the part I was retyping most.

3. Works where you already read. News, Wikipedia, Reddit, GitHub, papers, docs. Anywhere you can select text. There is also a popup launcher if you want to paste text into a blank tab.

What I would love feedback on:

  • Does ~25 free uses a month feel like a fair trial, or does it run out before you form a habit?

  • If you have tried inline AI extensions before and dropped them, what made you dropthem? That answer is worth more to me than a feature request.

posted toAvatar for product InstaPrompt
InstaPrompt
  1. 2

    The decision to not ask for an API key upfront is the most important product decision in here. That one choice removes the single biggest drop-off point in AI tool onboarding. I've watched teams kill products by front-loading API key setup -- most people won't do it if they haven't seen value yet. The 25 free uses question is the right one to be asking. The actual number matters less than whether 25 uses is enough for someone to build a workflow habit. Habits need repetition at the right frequency. If someone reads heavily and hits 25 uses in two days, that's the best outcome -- high engagement, natural conversion moment. If they hit 25 uses spread over three weeks, the habit may not have formed yet. Context Labels is the sleeper feature. Most people won't realize they want it until they've saved their first one and used it on a completely different site without having to think. That "oh, it just remembered" moment is when retention locks in. Do you let users create prompt templates with the label embedded, or is it a separate attachment that rides along? That distinction matters for whether power users build full workflows in it.

    1. 1
      Late reply, sorry, I was heads down on the relaunch. You read the API key decision exactly right. Version 1 asked for a key on the first screen and almost nobody got past it. Built-in AI was the single change that made installs turn into use. On your question: labels are separate objects, not text baked into a prompt. You save a label once (say, "grad student in biology, keep it technical"), then attach it to any prompt. The attachment sticks, so a prompt with its label is effectively a saved workflow, and the same label can ride along with five different prompts without being retyped. That is what I was retyping most, so it was the first thing I built after the key wall came down. On the 25 uses: your framing (two days versus three weeks) is the question I am now trying to answer with real usage data. [One sentence with what you actually see: for example "Most people who convert hit the limit inside the first week."] Happy to share what I learn.
  2. 2

    The “answer appears in place” detail is the part that makes this feel different. Keeping the context on the page removes a surprisingly annoying bit of friction, especially for people reading papers, docs, or long threads.

    This feels like it could be pulling a lot more organic traffic than it probably is right now, particularly with the no-API-key angle making the first try so easy.

    Have you noticed which kind of text people use it on most?

    1. 1
      Thanks, and sorry for the slow reply. I do not know per site, on purpose: the extension keeps no text and no page history, only a monthly token count. What I can see is prompts. Summarize and Explain lead by far, and the feedback points to long articles, papers and email drafts.
  3. 2

    The question about when users run out of free uses is interesting. That point could reveal quite a bit about how people actually use the extension.

    1. 1
      Agreed, and sorry for the late reply. Running out is the most honest signal I have: it means the habit formed.
      1. 1

        That “running out” signal is a strong proxy for habit. If you’re open to it, what’s the best email to reach you on?