Bootstrapping a Job Search AI Agent: Story behind LiftmyCV

Hey DEV.to,

I’d like to share the journey, challenges, and lessons learned while bootstrapping my startup/side project, LiftmyCV - an AI-powered tool designed to automate a job search and fill out applications on autopilot using ChatGPT (GPT-4).

The initial idea for my MVP was much simpler than an AI agent: a Chrome Extension paired with a web app.

How I started

Over a year ago, I was working for an early-stage startup that eventually declared bankruptcy. As a result, I began searching for new job opportunities. During the process, I realized how exhausting it was to search for jobs and fill out endless application forms across various platforms.

After some research, I found only a few tools that automated job searches and filled out information for me. However, they were expensive and full of bugs.

This gave me a green light to build something better with a more user-friendly business model - what I later called "Lifts." Of course, I also started considering ChatGPT for autofill features.

First steps

I didn’t (and still don’t) have enough coding experience, but I know a bit of PHP, WordPress, and UX/UI design from my time as a CPO and Product Manager.

Knowing that wasn’t enough to build such a complex tool, I decided to start small. I installed WordPress on my $5 /mo VPS hosting service and began as a no-code developer. I purchased a WordPress template and customized it, completing a pre-launch landing page in just 5 days.

At the same time, I designed wireframes and UX in Figma, finishing them in 2-3 days.

I decided to keep the landing pages on WordPress (at liftmycv.com) for better control over content, marketing, and admin tasks. Meanwhile, the core product would run as a Chrome Extension and web app on a subdomain (app.liftmycv.com).

Business model

I personally hate subscription models as a user but love them from a business standpoint. Plus, I’ve read posts and research showing that more users prefer pay-as-you-go models or one-time fees for lifetime access.

Since OpenAI API costs are high, I couldn’t offer a one-time fee. Instead, I created a pay-as-you-go model called "Lifts," where users buy Lifts for auto-apply features. One Lift equals one auto-apply.

To make LiftmyCV affordable, I optimized the first package to cost $15 - $20, significantly cheaper than competitors charging $100+.

First challenges

As I lacked the skills to build the automation tool myself, I spent a 1+ month searching for developers. I eventually hired a freelancer who agreed to build the MVP within three months.

My initial goal was to build an MVP in 3 - 4 months. Honestly, this was my third project where I thought I could build and launch in just three months. But here’s the reality: if you’re not a developer, it’s just not going to happen.

We set up a server on DigitalOcean, configured Git, integrated the OpenAI API, and began development. Meanwhile, I focused on legal matters and creating a Stripe dev account.

Next challenges

The developer seemed experienced, but after four months, I still didn’t have a working MVP. There were constant delays due to challenges with automation and other product features.

While waiting, I started learning about the OpenAI API and improving ChatGPT prompts. We faced issues with random responses during autofill, but after 4-5 iterations, we polished the results. Despite progress on ChatGPT integration and billing, the core automation tool remained buggy.

1 year later, I still didn’t feel like I could show this product (tool) to anyone or ask for feedback because it didn’t even work for me. Some might say you should start beta testing and launch earlier, but I believe beta testing only makes sense when your product is functional enough to solve the problem you set out to address. Plus, I didn’t think users would be willing to pay for a product that couldn’t even fulfill its main objective: auto-applying with AI.

Despair

I was ready to abandon the project because 1 year of waiting was too much, and that’s not how a startup should be built.

In parallel, I was listening to explanations from my developer about why we couldn’t do this or that. I have nothing bad to say about my developer — we’ve accomplished a lot together. We built the front and back-end, refined ChatGPT prompts and responses, and integrated billing. However, the core feature that inspired me to start this project wasn’t working as well as I’d hoped.

I knew automation was a tough part, but I also knew it was possible. What made me even more nervous was that during this 1-year delay, similar projects started popping up like mushrooms.

Turning point

I felt disappointed about all the time and effort I had invested into this project. I already had a fully implemented UI/UX design, billing, and even some users eager to test it out.

Despite everything, I decided to give it one last shot. I started searching for more developers and finally found someone who fixed all the problems within 14 days. 14 days, CARL! After waiting over a year, he managed to rewrite the entire automation process and Chrome Extension in just two weeks.

I was super excited and decided to start testing the product and automation on my own.

I realized the product was ready to be showcased to users when I received a few job interview invites within just 3 days of using it!

By the way, while writing this post, I got lucky and received my first production payment through Stripe. It’s a small milestone, but for early-stage founders like me, it’s incredibly motivating.

I haven’t even started pushing marketing yet — all I did was record a detailed YouTube demo and publish the Chrome extension to the Chrome Web Store. Before the launch, I even received a few investment offers, but I really want to keep LiftmyCV as a self-funded project and continue my bootstrapping journey. Anyway, let’s see what comes next..

My YT video that helped me get my first lead:

Next steps

I recently published the LiftmyCV Chrome Extension on the Chrome Web Store and switched from private beta to open beta. Now, anyone can register, test, and use the product. I’m also working on new WordPress landing pages to prepare for a global launch.

My long-term plan is to build a job search AI agent that can handle all aspects of the job search process, including filling out forms, preparing for interviews, sending follow-ups, and more.

Special offer for DEV.to

I don’t want to push self-promotion, but if you’re interested in testing LiftmyCV, your feedback would mean a lot. LiftmyCV offers a free trial with 3 Lifts to all users, but exclusively for DEV.to, I’ve created a promo code EARLY2025 for 25% off all packages. Unfortunately, I can’t offer completely free beta testing due to OpenAI/ChatGPT expenses. You can sign up using this link: https://www.liftmycv.com

If this isn’t what you’re looking for right now, I’d really appreciate it if you could share LiftmyCV with friends or colleagues who might be exploring new job opportunities.

Verdict

This is my 5th startup that I’ve bootstrapped from scratch, and 2.5 of them failed. The truth is, bootstrapping is especially challenging if you’re not a developer. It creates endless dependencies and time delays, and that’s why my biggest piece of advice is: start learning how to code as soon as possible.

Luckily, we now have tools like Cursor AI, AI Agents by OpenAI, and Claude. If you’re wondering why I didn’t start coding myself, the answer is simple: I was ready to invest my money to build something quickly by third-party developers while focusing on what I do best - product strategy and marketing.

But let’s face it - learning to code from scratch, even with tools like Cursor AI, would have taken me much longer. There’s a lot of new knowledge to learn from scratch. That said, once I establish a profitable SaaS business, I absolutely plan to start learning how to code using tools like Cursor AI or Claude.

Thanks to everyone who read this post all the way to the end! If you have any comments, questions, or feedback, feel free to share them in the comments below.

Cheers🍻
Dan