We (a tiny team of builders) just launched our chrome extension named ✨Tinker✨.

Tinker is a light-weight AI chat overlay to make your AI work-savvy.

We wanted to share it here first because this community understands the nuance of prompting better than anyone.

✨Tinker✨ – Website

✨Tinker✨ – Chrome Web Store

TL;DR

The real bottleneck isn't the prompt itself, it's the missing context.

The model is smart.

The prompt looks fine.

The answer is still mid.

Tinker sits inside any AI chat box (currently in ChatGPT / Claude / Gemini / Grok) and:

• Suggests 3 critical context tweaks in real time (like autocomplete, but for missing details).
• Lets you apply them with one click, instantly rewriting the prompt.
• Has a “One-Click Polish” button that infers missing context + cleans the prompt in one shot.

We think it’s a next level context-engineering tool on top of the classic AI chat interface.

The problem we’re obsessed with: “The Context Gap”

Everyone says “just talk to AI like a friend.”

In reality, it’s more like Slacking a busy colleague:

• They don’t see your screen
• They don’t know your boss
• They don’t know what “weekly report v2” means in your team

When we talk to humans, we naturally fill this gap:

“Hey, can you make a one-page summary for the VP, by tomorrow, bullet-pointed, focused on risks and next steps?”

With AI, people usually type:

“Summarize this.”

Same brain, less context.

We see that gap — between the messy intent in your head and the literal string the model receives — as the real bottleneck. That’s what Tinker tries to attack.

How it’s different from “prompt template” tools

We’re pretty anti–cookie-cutter mega templates.

Templates are great until:

• You’re staring at a giant form when you just wanted to “get this email out.”
• You’re copy-pasting “You are an expert X…” for the 40th time.

Instead of starting from a rigid structure, Tinker:

• Reads what you’re already typing
• Detects the biggest missing pieces of context
• Offers small, optional, inline nudges (like search autocomplete)
• Never blocks you with a modal or wizard

No new app. No second window. Just a thin “glass” layer on top of the chat box you already use.

Who we’re building for (aka: are you in this list?)

• Office workers / PMs / marketers who are tired of “meh” outputs from “Summarize this.”
• Creators who hate grinding prompts just to get the style right.
• Students / researchers juggling formal, casual, and analytical tones all in one day.
• Tech/product geeks who want a keyboard-first, inline, no-mouse, no-friction layer over all their AI tools.

If you’re the kind of person who already thinks in systems and prompt patterns, you’re probably the power user we want feedback from.

What we’d love from you🙏

If you’re up for it:

1. Try it on your real workflow
2. Tell us where the context suggestions suck.
   • Did Tinker ask for the wrong thing?
   • Were Tinker too timid and missed obvious gaps?
   • Did Tinker overdo it and annoy you?
3. Brutal takes welcome:
   • Is “Context Engineering” actually a thing or just new jargon?
   • What would make this actually indispensable for you?

We’re early — this is effectively v1 — but the mission is clear:

Make every person “AI-work-savvy” without forcing them to become full-time prompt engineers.

Happy to answer anything in the comments: tech stack, UX decisions, privacy concerns, roadmap (sliders for tone/length, macros/keyboard commands, etc.).

If you read this far, thank you 🙇‍♂️

Now please go bully our UX so we can make it better!