SkimIt.ai: How post-exit founders overcame PTSD to build a useful ai app
A personal struggle led to an actionable path for creators who feel stuck
I’ve had a hard time starting anything new since my last exit. After being stuck in founders block, I learned I could get back into my flow state by lowering the bar from “build the next unicorn” to “build something my friends use”. That change in mindset let Karthik and I overcome the Goldilocks Problem to build the entire SkimIt.ai app over MLK weekend.
In this post, I’ll go through:
The Goldilocks Problem and how it stops veteran founders from building
How I learned to lower the stakes
How we built hackathon guardrails to let us build unencumbered
How we came up with the SkimIt.ai concept (including a pic of our original whiteboard)
Movement over meditation: actionable takeaways from this experience
But first: what’s SkimIt? Simply email any article to firstname.lastname@example.org and we’ll email you back a TLDR summary of the article. SkimIt is perfect for all the Medium, Every.to, and Substack articles your friends send you but you’re too lazy to read. We’ll also send you drafts of tweets and LinkedIn posts so you can sound super smart in front of your friends. We have not marketed the app at all. Yet somehow we have 170+ people actively using it… I only recognize 5 of them. Super fun watching it grow organically. [Update: over 900 unique people have used this tool]
I struggled to begin new projects since selling my last company. Even though my PTSD subsided, I could not forget the difficulty of building a real business. Without the childish hubris of my 20s, I was stuck in the Goldilocks Paradox:
Goldilocks Pardox: Any new startup idea was either too difficult or too inconsequential. No idea is perfectly worth diving into.
After going through the entrepreneurial gauntlet twice, my bar for the “right idea” became unreachable. Some ideas were easy to execute but too small; they would never become the massively impactful company to which I’d want to dedicate the next decade of my life. Other ideas were obvious unicorns but felt completely out of my reach; how am I supposed to learn genomics to create an ai bioreactor platform??
I was stuck — staring at a page of scribbled ideas but unable to enter my flow state of building. I love building, but I could not waste time on any suboptimal ideas.
The Veteran Founder Trap
It turns out I’m not alone. I recently joined a post-exit founder group with dozens of similarly stuck founders. We literally held a 20-person zoom session that amounted to group therapy.
My friend Karthik was going through his own post-exit journey. He is a natural builder: super sharp, a fast learner, and relentlessly motivated. He’s a beast. He built an incredible company (Kinnek) but had not coded much since his exit about a year earlier.
Last year, he and I had spoken on and off about building something together. We even tried working on a concept last year but dropped the idea after spending three weeks putting it through the wringer and realizing it wasn’t good enough (see Goldilocks Problem above!)
Revelation: Lower the stakes to just play
My big revelation came late last year — I felt my inner playful child reborn and resolved to lower the stakes for myself. I would play with software the way I used to play with LEGOs. I gained faith that by building and learning in a space I love, my progress will lead me to naturally start a company or join a team. Rather than building to create a massive enterprise, I would build because I loved creating products for people.
This liberating mentality let me dive into generative ai and build Sage: the AI Therapist chatbot modeled after Esther Perel at the end of 2022. I published the story of Sage last week, and already I’m getting inbounds from talented people who want to collab.
Solution: The time-boxed hackathon
For Q1 2023, my goal is to build alongside highly-talented engineers. I am structuring each build as a time-boxed hackathon: we have 4-7 days to take an idea from concept to deliverable.
Instead of building long-term stakeholder value, we focus on creating something fun and approachable in an intellectually stimulating space.
We rewired our brains to remember the most fun part of building a company — that initial spark of ideation that comes to life as a product and brings value to real people.
Creating these guardrails stops us from overthinking everything. We don’t get fixated on fundraising decks, exit multiples, customer acquisition funnels, risk management, or staffing.
Thanks for reading Magnetic! Subscribe to receive new posts and support my work.
How the SkimIt Ai app came to life
Karthik and I teamed up for the first hackathon. We met at Nomadworks NYC, bounced around a few ideas, and did prelim research on each to rate them on two qualifiers:
Can we put this together within one weekend?
Would we or anyone we know actually use the thing we build?
Karthik suggested the concept of an ai-powered daily digest. The initial concept involved a presidential briefing delivered to your inbox every morning that summarized the various information thrown at you from all your sources. From there, we agreed that the most profound technical challenge was the automated summarization of information into an easily digestible format.
The hard time limits forced us to continually reduce scope:
Creating a fully baked app where you would connect your Gmail and we’d crawl through all of your emails? Too hard! Users will simply email us the content they want summarized.
Summarizing an entire day’s information flow? Too hard! Start with one article.
The reduced scope led to unexpected benefits:
Users interacting with the app via email naturally solves user authentication. There’s no need to have users log in. This is similar to how my Sage AI chatbot authenticated users through their phone numbers.
Focuseing on one article at a time lets us give additional value by pre-drafting tweets and social posts about that content.
Lesson learned: timeboxing ourselves with the long weekend kept us focused on practical features for real users
Building the SkimIt user flow
Here’s a pic of our initial whiteboarding for SkimIt:
We stuck tightly to our initial UX goals in the left column and even managed to ship the first of the “Possible Iterations” on the right column.
The Finished Product
Feel free to check out www.skimit.ai and let us know what you think! Just send an email to email@example.com with a link to any article in the body of the email.
Here’s a screenshot of a summary email you’d get from SkimIt. Personally, I love the tweets it drafts :)
Movement Over Meditation
“Look inward.” “Do some soul searching.” “Take time to find yourself.”
I might be living in social bubble, but these are the tropes I’d hear from people trying to steer me out of my funk. This might work for others, but my experience has been to favor movement over meditation.
Putting one foot in front of the other in any positive direction is better than endlessly pondering which way to walk.
Movement creates momentum. Even if building an ai summarizer will not define my next decade, I:
Rediscovered my flow state of building
Affirmed to my ego that “I still got it!”
Realized ai and python come somewhat naturally to me
Have a portfolio of stimulating projects to share with my network! I’m no longer the guy at the dinner party who is “between projects”; I’m now the guy who is building in ai. I believe that this momentum will lead me toward the right project, co-founder, or team.
Lessons learned for the next hackathon
We should have done a structured daily standup to share updates and set goals; we both felt overly motivated to perform well in front of each other… but without goalposts, I always felt like I wasn’t doing enough. Good for motivation but not ideal for the psyche.
Since we split up the work, we each became experts in our respective silos; after finishing the project, we should set aside time to teach each other everything we learned along the way.
We should have spun up the dev —> test —> deploy pipeline on our local devices and Heroku first and foremost. Instead, we had two separate codebases on our devices, and Karthik had to merge my code into his. I was too scared to break the deployment pipeline by committing to the repo. The longer you delay it, the more of a pain in the ass DevOps becomes
Overall, I’m beyond grateful to Karthik for building this fun app with me. I’m confident he will do something incredible!