From 0 to MVP
A Guide to Building Your First Minimum Viable Product with AI
Hey builders!
I’m excited to share a piece I’ve been working on for a minute. A lot has happened since you last heard from me and we’ve got a lot of new subscribers! So before diving in, I wanted to thank the newcomers for joining along this journey!
The purpose of My AI Co-Founder is to document the rise of AI coding and software-building among the non-technical solo builders who I believe are some of the best positioned to benefit most from the AI revolution.
This is the final post in a 3-part intro series to help anyone, regardless of technical ability, understand why AI is helpful and start using it to build software today:
Part 1: The New Solo Founder's Journey
Overview of how non-technical solo builders can benefit from AI and why I started My AI Co-Founder to help evangelize these benefits.
Part 2: Getting Started with AI
A tactical overview to demystify and familiarize with AI coding and start using AI to build basic software.
Part 3: Build an MVP ← This post
A step-by-step guide and tool overview to build a minimum viable product for your initial users.
If you’re non-technical and want to build (or are already building) with AI and want a community of people doing the same, join our bi-weekly AI Office Hours! The next one is Monday @ 8:30am PST, you can RSVP here.
Alright, let’s dive into MVPs
Make sure to read to the end for a sneak peak of the direction I’m planning on taking MAIC (hint: it might involve you!)
"If you're not embarrassed by the first version of your product, you've launched too late." - Reid Hoffman, co-founder of LinkedIn & PayPal
It’s reassuring that even billionaire founders like Reid Hoffman were embarrassed by their product’s first version. I was with Contxt and you should be with yours.
The first version of your build is called your MVP, or Minimum Viable Product. Eric Ries popularized this term in his seminal book, The Lean Startup, which evangelized the Build-Measure-Learn Feedback loop.
This loop encourages entrepreneurs to build an MVP, measure its market performance, and quickly learn from the results to decide whether to pivot or persevere. This has become the gold standard of product-building in Silicon Valley and it’s the methodology I swear by and will encourage all builders to practice here at MAIC.
Building an MVP is crucial because it reinforces building quickly and integrating insights from real users. This framework helps you launch quicker, mitigate risk by conserving resources, and increase the odds of entering the market with a product that solves real problems (achieving Product Market Fit).
What’s your MVP?
How do you know what your MVP is?
Ries summarizes an MVP as “the smallest thing you can make or do to test your hypothesis.” But, what’s your hypothesis? And how can you test it quickly?
In short, your hypothesis is the problem you’re trying to solve, and you’re searching for proof that your solution is the right one. You should search for proof of two things:
You’re solving a real problem for real people.
That your solution actually solves that problem.
Michael Seibel, Director of Y Combinator & co-founder of Twitch, summarizes the true goal of an MVP as “the first thing you can give to the very first set of users you wanna target, in order to see if you can deliver any value at all to them.” We need to get out of the fantasy worlds in our heads and get practical about what we can make happen in the shortest amount of time.
Y Combinator is my go-to source for high-signal, vetted startup guidance. The quote is from their How to Plan an MVP video, which I’d recommend pairing with How to Build an MVP.
MVP Case Study
Let’s use the Contxt MVP as an example.
In Nov 2023, I needed to figure out where to start amidst the exciting ideas I have for Contxt. So I asked myself 3 questions to clarify my hypothesis and focus:
What’s the first problem I’m trying to solve?
Who else experiences this problem?
What are the 1-2 things I need to build to “deliver any value at all” to the people experiencing this problem?
Problem
I can’t easily save and search quotes from my friends.
I’d been thinking about it for years and this was the most urgent problem I could solve for myself and see if it resonated with people who did the same quote-keeping practice as me. There were many problems I highlighted, but chose not to address with my MVP:
I can’t easily filter or sort the quotes I keep.
I can’t easily share my quotes on social media.
I can’t easily add privacy settings for each quote.
I can’t easily store a quote from a text message exchange.
I don’t have fun ways to rediscover old quotes.
I couldn’t build an MVP for all these problems, so I chose to start with the one I suspected most Quote Keepers like myself would share and that I could address soonest.
Who has this problem?
After years of showing people quotes stored in the notes section of my Apple Contacts, I knew there was a small contingent of individuals I affectionately came to call Quote Keepers: people who enjoy writing down their friends’ quotes.
To identify who shares the same problem you’re trying to solve, I highly recommend reading Lenny’s post on finding your Ideal Customer Profile. He explains how to think about your initial target customer and shares tactical advice on identifying them via hyper-specific characteristics. Here were the three I identified for Contxt:
The ideal Contxt user:
Kept a quote in the past week
This ensures I’m focused on people who would use Contxt daily/weekly.
Uses Apple Notes, Google Docs, or Notion for storing quotes on their phone.
Many quote keepers store quotes in journals, but I’m building an app, so I’m focused on tech-native people open to a hyper-specific solution to replace the general tools they’re using on mobile.
Prioritizes IRL socializing
The best quotes happen IRL, and I’m building Quote Keepers that feel the same and live accordingly.
What to build
There are two parts to the problem I wanted to solve with an MVP.
First, easily save quotes. Across all the platforms I’ve used, getting to a contact, adding a section for a new quote, and writing the quote was annoying, inefficient, and (in Apple Contacts’ case) buggy. Making it easy to start typing a quote was essential.
Second, easily search quotes. In Notion and Apple Contacts, I couldn’t locate the quotes I created. I wanted search to surface the exact quote in the friend’s profile.
The Contxt MVP
Starting December 2023, I focused on Contxt every morning and was able to launch the MVP in March 2024!
I was excited but crucially, I was also embarrassed. The app was buggy as fuck and barely got the job done. But I could show something I built to the world, address problems I saw with fellow Quote Keepers, see if it worked well enough, and then iterate based on feedback.
Side Note on AI at this Stage
You might’ve noticed I’m not encouraging the use of AI to answer these questions.
That’s intentional, as I believe product-building is a pseudo-spiritual practice that merges practical thinking, personal intuition, internal reflection, and interactions with the external world. These personal considerations deeply impact your product and are best honed by reflecting and inquiring into yourself, other humans, and ideally, your future users.
If you’re stumped by the questions above, feel free to brainstorm and ask Claude/ChatGPT. AI can challenge your assumptions or act as a sounding board for these questions. But it doesn’t know your lived experience and only seeks to please you with its answers, not think critically about the nuances.
Building your MVP
After planning your MVP, decide whether to build a website or a mobile app. Web is easier to start with and improve over time, but Mobile offers unparalleled functionality hard to recreate on a website.
Once you choose, skip to the relevant platform section below.
Websites
To quickly set up a web application, choose between DIY or full-stack coding. Both have strengths and weaknesses:
Approach 1: DIY
This approach is my personal recommendation because it’s the new way of building software and will only get easier. New walkthroughs and videos are shared weekly on Twitter, TikTok, and YouTube showing non-technical people how to use the AI tools below.
If you want custom code and to build apps like professional developers, this is the approach for you.
Tools
Cursor (Code Editor) - This is my preferred code editor. It’s challenging Microsoft’s VS Pro and is AI-native. It’s quickly becoming the favorite for AI-native developers.
I attended a Cursor meetup in SF two weeks ago and learned their entire 20+ person team is comprised of engineers. No HR, Marketing, Sales, or Developer Relations, so they’re 100% focused on building the best product possible, which is promising for a developer-focused tool.
Here’s a great walk-through of how to pair Replit and Cursor.
Claude (AI Writing & Coding) - The best AI tool for questions and rapid prototypes. Read my previous article to learn more about it and why I love it.
Replit (AI Coding & Deployment) - This is the best and easiest platform to deploy a website. With Replit’s AI Agent release last week, it’s now ridiculously easy to have AI build and iterate on your product (more on this below).
Firebase (Backend) - While I’d prefer a non-Google backend service, Firebase is a great starting point. It’s a Google-backed Backend as a Service (BaaS) platform for building, testing, releasing, and monitoring web and mobile apps.
Walkthroughs
Once you have these tools, check the videos below for the full process of building and launching a website:
Last week, Replit announced their new AI Agent functionality which allows their AI to plan and execute tasks independently. It’s mind-blowing and enables unparalleled non-technical execution that I’d highly encourage you to explore.
Checkout this walkthrough for setting up a start-up waitlist landing page.
Checkout this Twitter thread with additional examples of what’s possible.
Replit’s walkthrough (Claude + Replit)
Riley’s walkthrough (Claude + Replit + Firebase)
These are great starting points, but leave a comment if you want more resources and I’ll share them!
Approach 1: Language-based
The apps below handle coding, deployment, and database creation. They use natural language to build fully functional apps, making it fast and efficient for non-technical users.
If you want to get up and running quickly without navigating multiple apps and interfaces, this is the approach for you.
Tools
A free platform for rapid development and deployment of full-stack applications using Next.js, enabling quick creation of complex apps while facilitating iteration and bug fixing. They offer an Odyssey Program to help new builders get their first 1,000 users in 4 weeks, which is launching in just 5 days!
A free AI app builder where you describe your desired application, and the platform converts it into code. It offers various style guides, enabling users to publish their sites or tools without templates, making it ideal for quickly bringing ideas to life.
These apps deploy a full stack that can be pushed to GitHub and fully edited in an IDE like Cursor.
Mobile
Building a native mobile app is different from building a website. Mobile apps are distributed through app stores and require user downloads for updates, while websites are instantly accessible via browsers (which makes it easier to share product changes/fixes). The complexity of developing for multiple platforms (Apple and Google) and maintaining compatibility, along with the app store approval process, can make mobile app development more nuanced and challenging.
A simple workaround many apps use is to build a mobile-optimized website and have users add the URL as an app icon. While this works, this approach doesn’t provide access to native UI components for a seamless, platform-specific user experience that takes advantage of all the device features that browsers don’t have access to.
If you want to build the best mobile experience, this is the approach for you.
Approaches
DraftBit (React)
Draftbit lets you design, build, test, and publish React Native mobile apps from one place. You can export your project in React Native code to continue building in your own environment. I used DraftBit to build my MVP for Contxt and had a great experience with their Expert service to get from 0-1 when AI wasn’t good enough last year.
Flutterflow (Flutter)
Flutterflow is like DraftBit, but their apps are exported in Flutter code, not React.
I won’t go into the differences between Flutter and React, but Flutter provides a streamlined experience with consistent visuals across platforms, making it easier to design unique interfaces without deep technical knowledge. React Native adapts to look native on different devices, which is advantageous if you want your app to use system UI for Android/iOS. Your choice is between a consistent-looking app (Flutter) or one that adapts its appearance to match each platform (React Native).
Native (Swift for iOS or Kotlin for Android)
If you want to build outside an all-in-one platform, or want to focus solely on iOS (Swift) or Android (Kotlin), consider native development. After building the Contxt MVP in React, this is the approach I’ll be taking moving forward with Contxt.
I’m an iPhone user, so while React and Flutter let me create an app for both iOS and Android, I want to build in Swift to leverage Apple’s simpler development process, avoid library incompatibilities, and access upcoming Apple Intelligence AI features.
While React/Flutter help you deploy across all major platforms, that distribution efficiency comes with the cost of complex code libraries and frameworks to enable one code-base to work on iOS and Android. These libraries/frameworks are constantly updated and often cause compatibility issues that I’ve been trying to resolve since exporting my code from DraftBit. These issues are hard to correct with current AI, which is a major reason why I’m going to develop in Swift.
Concluding Thoughts
Building an MVP isn't about perfection – it's about progress.
Whether you code a website from scratch or use no-code tools for a mobile app, the important thing is to start. I was embarrassed by Contxt's first version, as was Reid Hoffman for LinkedIn and PayPal. Your MVP is your learning tool; it'll help you figure out if you're solving a real problem for real people, which is the only thing that really matters.
Choose your approach, set a deadline, and get your idea out there. Don’t worry about making it flawless – it won’t be, just make it exist. The sooner you launch, the sooner you’ll learn. Those initial learnings and connections with early adopters are worth their weight in gold and will launch you in directions you never would’ve otherwise thought of
Housekeeping
Before you go, I have a few quick housekeeping things to share. If any of them resonate with you and you’d like to learn more, please email me!
Are you a builder?
I’m looking to build a collective of builders to document and showcase the process of building AI-enabled software.
My plan with MAIC is to have different product tracks based on where people are building (Web, Cross-platform, iOS, Android, etc.) that newcomers can follow to get started. If you’re pursuing any of these paths, I’d love to chat and see how I can support and document your journey so others can learn what you’re learning 🙌
Feedback - If you have thoughts or feedback about MAIC and what you’d like to see, please leave a comment or email me. I’m looking for ways to improve and relevant topics to address, and I read every email and comment.
Know someone who you think would be interested in MAIC?
Tap the button below and you’ll get a link to share on socials or in the chat!