Inside the Pod: Momentum Is the Key to Productivity

Thank you to everyone who is watching or listening to my podcast, AI & I. If you want to see a collection of all of the prompts and responses in one place, Every contributor Rhea Purohit is breaking them down for you to replicate. Let us know what else you’d like to see in these guides.—Dan Shipper

Was this newsletter forwarded to you? Sign up to get it in your inbox.

---

Have you ever opened up your laptop after dinner just to reply to that one email?

And then you end up replying to the email, drafting three more that you’ve been putting off, and getting a headstart on prep for that meeting you have tomorrow. Not before long you’re staring at your screen bleary-eyed, wondering how you got sucked into working until midnight.

There’s a simple explanation for this: momentum.

To be clear, working late into the wee hours of the morning is not advisable. But that doesn’t take away from the fact that momentum is a powerful thing. If you can get the proverbial ball rolling, the odds are you can keep it rolling. 

Momentum is the driving force behind Grimoire, the top-ranked custom GPT for programming created by Nick Dobos. 

Once you log onto Grimoire, you are presented with a selection of projects you can get started on right away, like a tool to create a link-in-bio website. And once you start working on a project in Grimoire, Dobos keeps the momentum going in two ways: one, by getting the GPT to provide a step-by-step outline of what is needed to complete your project; and two, programming it to recognize “hotkeys” or specific letters on the keyboard that, when pressed, trigger the GPT to execute a pre-programmed function. 

Dobos has designed Grimoire to propel users forward. If that weren’t enough, Grimoire boasts a whimsical fantasy theme. The coding assistant is peppered with references to magical creatures, transforming the experience into something akin to attending a class at Hogwarts.

In this episode of AI & I, Dan Shipper interviewed Dobos to talk to him about Grimoire, his insights from building one of the most popular programming custom GPTs, and how AI is catalyzing the rise of a new paradigm in coding.

In this essay, I’ll pull out the core themes of their conversation (along with accompanying screenshots of their chats with Grimoire!):

• How to ship a website with a two-word prompt and a single click
• The best way to learn how to code in the age of AI
• What gives Grimoire an edge over ChatGPT

AI fundamentally changes the way humans interact with computers—and I think these tips will be interesting to anyone who wants to leverage this to their advantage.

How to ship a website with a two-word prompt and a single click

Dobos thinks the sheer scope of what one can do with AI sometimes overwhelms people into inaction. In order to solve the “blank canvas problem,” Dobos has built “templates” on Grimoire that allow people to jump straight into a project.

All screenshots courtesy of AI & I.

Dobos notes that these templates are merely starting points, and do not limit users from prompting Grimoire to execute their unique ideas. Inspired by the cup of coffee on his desk, Dobos decides to use the custom GPT to create a website about coffee.

Dobos: Coffee website 

Given just a two-word prompt, Grimoire generates a comprehensive plan for the website's structure and deployment. It also presents the user with a series of next steps, offering a list of hotkeys. Here’s part of what Grimoire generated:

The hotkeys are among Dobos’s favorite features on Grimoire. He explains that if a user inputs the letter “K” into a chat with the GPT, it will display the full array of hotkeys available. Dobos wants to demonstrate how easy AI has made it to execute projects like making a website, so he uses the hotkey “N” that pushes the website live nearly instantly. 

Dobos: N

While Grimoire processes the prompt further, Dobos elaborates on its underlying mechanics. He explains that Grimoire is partnered with Netlify, a hosting provider that enables instant website deployment. The generated website is temporary, set to self-delete after an hour, unless a user chooses to preserve it by signing up with Netlify and paying a hosting fee. As Dobos talks, the custom GPT has collaborated with the hosting provider, and the coffee website is ready. He clicks on the URL to visit the site.

Dobos candidly acknowledges that the result isn't a masterpiece of web design. However, he emphasizes the remarkable feat it represents: an entire website conjured from just a two-word prompt and a single click. Taking it a step further, Dobos prompts Grimoire to create a shareable QR code of the coffee website’s URL in the same chat.

Dobos: Let’s make a QR code for the website URL

Watching Grimoire is like watching a very focused intern hard at work. The GPT methodically progresses through a series of steps: First, it formulates a plan for code generation; next, it implements this plan using Python; finally, it executes the code, producing a link where users can access and download the QR code.

Dobos clicks on the link—and the shareable QR code for the Coffee Lovers website that Grimoire created is ready!

The best way to learn how to code in the age of AI

Another project that Dobos has built into Grimoire is a comprehensive curriculum designed to teach people how to code from scratch. He recognizes a command dilemma faced by aspiring coders in the AI age: whether to learn programming traditionally before leveraging AI tools, or to piggyback on AI assistance from the outset. Dobos sees this uncertainty as a barrier for many who understand the value of coding skills, but are unsure how to approach how to learn it. He hopes that his curriculum addresses this gap. 

Dobos has programmed a hotkey, the letter “P,” to generate this curriculum in Grimoire. 

Dobos: P

Dobos has infused his coding module with a whimsical fantasy theme, almost recreating the atmosphere of a lesson at Hogwarts. The curriculum cleverly disguises programming concepts as magical instruction, starting with "Beginner's Incantations" for basic concepts and progressing to "Forbidden Spells" for complex techniques. 

While discussing his broader mission around the curriculum, Nick explains that creating the module with ChatGPT helped him articulate his knowledge in a useful manner. “Can I take a big part of my expert knowledge that I've built up over the years and turn that into a course, give that back to the community, and do that in a weird interactive chatbot?” is a question he hopes the curriculum has resolved.

What gives Grimoire an edge over ChatGPT

Dobos thinks that Grimoire has an edge over the regular version of ChatGPT for the following reasons:

• Baked-in custom instructions for better performance. Dobos observed that LLMs perform optimally when given specific guidance—for instance, if you prompt an LLM to respond as an expert, it’s likely to give you a much better answer than if you didn’t include that in your instructions. To that extent, while building Grimoire, Dobos incorporates tailored instructions that “bump [the model] into being more truthful, more helpful, more thoughtful.”
• Blueprint before you build. Grimoire is programmed to prioritize project planning. When given a prompt, the GPT first outlines the preparatory steps required before moving to execution. Dobos believes this feature enhances a user’s ability to plan effectively and approach projects in a structured manner.
• Hotkeys that prime you to act. Dobos thinks that the hotkeys integrated into Grimoire act as subtle prompts for users, encouraging them to proactively think about what their next steps in the project should be.  

To demonstrate these qualities, Dobos prompts Grimoire to address one of the random programming questions he often finds himself asking the model.

Dobos: How do I make a game in Swift?

For context, Swift is a programming language designed by Apple to create apps for iPhones and iPads, and something Dobos uses often as an iOS programmer. Grimoire, just as Dobos expected, outlined a step-by-step approach to address the query. Here’s part of what the model generated:

Dobos remarks that one of the reasons he included hotkeys in Grimoire is because he wanted a quick way to execute desired actions without having to type a lot. To demonstrate this feature, he showcases one of his frequently used hotkeys, the letter “Z.” This particular shortcut automates the process of saving and downloading the code generated by the model into a designated folder. 

Dobos: Z

As Grimoire generates a response, Dobos explains that the hotkeys also help him avoid going “down a rabbit hole, [getting] weird answers that make no sense and run in circles.”

During the course of this episode, Dan and Dobos deployed a real website and got instant answers about a specific programming question—without expending significant time or effort. This new paradigm is only possible because of the capabilities of AI. This can be overwhelming, but lucky for us, Dobos has also designed a curriculum to help those who want to learn how to navigate it. All this to say, the new language of programming is here, and if you’re reading this, you’re already fluent.   

---

Rhea Purohit is a contributing writer for Every focused on research-driven storytelling in tech. You can follow her on X at @RheaPurohit1 and on LinkedIn, and Every on X at @every and on LinkedIn.