Have you ever ever had a terrific idea for a product, but weren’t sure if there was enough demand for it or if you happen to had enough time to develop it?
That was the situation I discovered myself in while attending a conference with my company and now, two weeks later, we’re able to launch R.Ai on Product Hunt.
It was April thirteenth, and I used to be attending the AACR Annual meeting, certainly one of the most important cancer conferences on the planet. We had a booth within the exhibition, so we arrived just a few days early to establish and ensure the whole lot was so as. In the course of the downtime between talks and commutes, our CEO shared an concept that modified the whole lot.
“Wouldn’t or not it’s great if I could just stop having to put in writing so many emails? That’s something I’d pay for,” he said.
“That’s something I can construct,” I replied.
“We shouldn’t construct that if we are able to’t sell it,” he countered.
At this point, it’s price noting that we’re a biotech company specialising in single-cell RNA sequencing evaluation. As a pc scientist, constructing an AI email assistant will not be a part of my typical workload. Nonetheless, I had been happening endlessly about how easy it’s to construct prototypes with ChatGPT, and I believed, why not give it a try?
The challenge was to confirm if there was enough market demand for our Gmail AI assistant. As you’ve probably read elsewhere, your goal is to be sure that someone is willing to pay on your product without having to completely develop it (I believe I read that in the Lean Startup). Essentially the most common way is to develop an initial version, called minimum viable product (MVP), and check out to sell it to see if anyone is definitely interested. This manner you’ll be able to validate your idea and getting early user feedback without committing all of the resources needed to construct the complete product. Since I had some spare time at our booth, I made a decision to present it a go and take step one.
I started by deciding on the sort of product I wanted to construct after which tried to quickly implement an MVP for the front-end part.
As a newcomer to Gmail plugin development, I turned to ChatGPT for guidance. Crafting effective prompts will be difficult, but my goal was to receive a comprehensive list of options after which choose the perfect approach. So, I posed the next query:
> Because the principal engineer of a latest company, we wish to supply an AI system that auto-generates replies to emails. We predict it must be integrated into Gmail, but we’re unsure about the perfect technical option. Are you able to provide suggestions and evaluate them, considering we wish something fast and easy for the demo and MVP?
ChatGPT provided a radical evaluation, presenting the professionals and cons of making a browser extension, a standalone web app, and a Gmail add-on.
When working with ChatGPT, it’s helpful to request high-level steps to finish a task, allowing you to concentrate on one step at a time while maintaining a relevant context. Remember that ChatGPT may make assumptions that don’t suit your project, so it’s essential to make clear them early on. In my case, I asked for a suggestion based on ease of implementation, a top level view of the required work, and an outline of the user interaction.
> Select certainly one of the choices based on how easy it’s to implement, and supply a top level view of the work to do. Also, give a temporary description of how the user would interact with the feature.
This prompt proved useful, because it included deciding which browser to make use of and the back-end implementation within the work plan. Wanting to focus solely on the front-end, I specified that we’d use Chrome and that it could assume that the back-end interface could be a single endpoint. How one can actually construct such a back-end I’ll inform you in a later post. With these details ironed out, we advanced to the subsequent phase.
With a transparent idea of what I needed to create, it was time to dive into the actual development process. My prompt for ChatGPT was:
> We’ll use Google Chrome. Assume the back-end will likely be an endpoint where you do a post request with the e-mail text and receive a text answer. The user interaction will consist of a single button that generates a whole reply, which users can then edit. Now let’s start coding — are you able to explain the way you’ll develop the feature?
ChatGPT responded with a comprehensive guide on implementing and testing the browser extension. Here’s a temporary rundown of the steps, excluding the actual file contents:
- Create a latest folder for the extension and arrange the crucial files:
a.manifest.json: Accommodates metadata in regards to the extension, equivalent to its name, version, and permissions.
b.content_script.js: Injects custom functionality into Gmail’s web interface.
c.background.js: Manages background tasks, like communication between the content script and your back-end server. - Define the extension’s metadata within the
manifest.jsonfile. - Implement the content script in
content_script.js. - Implement the background script in
background.js. - Test the extension locally:
a. Open Google Chrome and navigate tochrome://extensions.
b. Enable “Developer mode” within the top-right corner.
c. Click “Load unpacked” and choose the folder containing your extension files. The extension should now appear within the list and be enabled.
After requesting the contents of popup.html and popup.js, I put together the files — manifest.json, content_script.js, popup.html, and popup.js — and decided to try loading it into Chrome. To my surprise, it worked straight away. I had a working Chrome Extension very quickly without even reading the code! Really wow.
Up up to now browser extensions were something complicated that only really purposeful and dedicated people (like Henrik Aasted Sørensen, the guy from Adblock Plus, who made browsing bearable again) would attempt, not lazy me. But that was all and in lower than half day I had a working Chrome extension (kudos to google also for the way easy to check this out in Chrome is).
after getting the extension to work, I lost half a day asking chatGPT to craft different CSS files for it within the type of Bitwarden, Material Design, or AdBlock Plus. It is very easy. So be ware of getting distracted with the numerous random things ChatGPT can do for you if focus is to quickly iterate.
So there you might have it! With ChatGPT on my side, I managed to create the primary version of my extension in only a day. It goes to point out that turning an idea right into a product doesn’t need to be an enormous ordeal anymore (and lots of committed weekends). With the precise AI tools, like ChatGPT, you’ll be able to bring your ideas to life faster than ever before.
Now, my next move is to establish a basic landing page to showcase the product and gather some precious feedback on its potential market demand. Keep a watch out for my next post, where I’ll take you thru the strategy of constructing the web site, adding Google sign-in, integrating Stripe payments, and sharing useful recommendations on easy methods to turn your personal ideas right into a reality. Whether the ideas are good still is determined by you 😉
And hey, don’t forget that the perfect is to maintain iterating and improving. Should you’ve got any suggestions to make this post more interesting or engaging, don’t be shy — let me know! and if you happen to liked the product try R.Ai on Product Hunt!