This essay as told is based on a conversation with Julia Starr, the founder of Julia Starr Coaching. It has been edited for length and clarity.
I’ve been a career coach for almost a decade, and for most of that time, my business has been capped by one factor: me. As a solopreneur, I have limited space on my calendar and I’ve never been interested in growing a team.
About a year and a half ago I started experimenting with how AI could help me develop my impact. I created prompts that clients could use to deepen their knowledge between our sessions, and I created an AI “co-coach” to support participants throughout each week of my group coaching program.
Last year, I took the most transformational step yet: I vibe coded an app which reproduces my basic methodology. This provides me with passive income and expands access to my work.
I tried vibe coding without preparation
In December, I signed up for a hackathon with Kindan AI tool that allows non-technical people build apps using natural language prompts. I decided to really dedicate myself to it, clearing my calendar and waking up at 5 a.m. to take part in the kickoff calls.
Starr had no coding knowledge when she signed up for an ambient coding hackathon. Robert Warren for BI
I was walking in with literally zero coding or product knowledge. What I had was my brand coaching framework: the VSA method, where clients dig deep into their core values and strengths before taking small steps to prototype different paths forward. I was wondering if I could put this in an app.
I started by saying ChatGPT I wanted to code a product and asked where to start. Next, I studied my prompts in ChatGPT before integrating them into Lovable.
After two days of this type of back and forth, I had a starter version of the app. I tested it in the Lovable community, and pretty immediately people came back saying it gave them useful ideas they never would have thought of on their own. I was surprised.
The initial result was impressive, but needed refinement
The app was promising, but it wasn’t perfect yet. I spent the next month building it into a product that worked consistently.
User testing helped Starr refine the experience and cost of its app. Robert Warren for BI
For example, it regularly happens crazy job titles it didn’t exist. I thought I could just tell Lovable that the information must be legitimate and that it would be corrected, but that didn’t work. I had to be creative and precise in setting the quality bar, so I asked Lovable to find 100 instances of a particular job title on LinkedIn before surfacing it.
I also did real user tests. I would ask people to share their screens and browse my app while speaking out loud their inner dialogue – what seemed confusing to them, what surprised them, what they wanted next. I recorded everything, then dumped those transcriptions into Claude to identify key changes to make and how to write the next prompt for Lovable. I did this after every conversation.
After approximately 50 to 60 additional hours of work beyond the hackathonI publicly launched my app, Threshold, in January.
My AI app provides passive income
I originally wanted to charge over $150 for the product because my intellectual property is valuable. However, during user testing, people told me they wouldn’t pay that much unless they already knew me and my work.
Starr said she isn’t worried about AI replacing her coaching expertise. Robert Warren for BI
Based on this feedback, I decided that the app would be more valuable if it served as a marketing funnel this could drive new customers to my more expensive offerings, rather than charging extra for each individual app sale.
And I never worried that the app would replace me entirely; I’ve been coaching long enough to know that having information doesn’t automatically lead to change. There’s a real gap between knowing what you could do and actually doing it, and bridging that gap usually requires a trusting human connection.
Nonetheless, I wanted to put up a paywall to ensure that customers take their career pivotsso I had to pay a one-time payment of $29. At this price point, I’ve found that people share my app organically and 75% of those who hit the paywall convert, so I’m earning significant passive income. I make small adjustments based on feedback here and there, but it’s mostly hands-off now.
What excites me most is what my app makes possible in terms of access. There are so many people who will never be able to work directly with me due to time or money constraints. This app can help them get started – and for someone who’s stuck, that first unlock changes everything.