How AI Is Changing Operations for Outdoor Brands and Why Every Company Can Build Software Now
Jason Kuperberg on viral “Roast My Strava,” a practical AI adoption playbook, and angel investing
Building software used to require teams of engineers and serious capital. Jason Kuperberg built Roast My Strava in a few hours from the back seat of a car, and it took off because people genuinely wanted to share it.
That story isn’t just a fun flex. It’s a signal: the cost (and friction) of creating digital experiences has collapsed. For outdoor brands, that changes the playbook. AI is not only about saving time on copy and ops. It’s also a new path to interactive experiences people actually want to use and pass around.
🎧 Listen to / watch the episode: Spotify | Apple Podcasts | YouTube
From early AI experiments to millions of users
Jason got deep into generative AI in 2020, before ChatGPT made large language models a mainstream household concept. He and his co-founder Matt were testing early models that could string together coherent text and code. They saw the curve: capability was compounding fast.
They posted demos of AI writing emails and code. Investors reached out. That momentum turned into Other Side AI and their product HyperWrite, which Jason says now has millions of users.
But the post-ChatGPT gold rush changed the market. When everyone is “an AI company,” differentiation gets harder and burnout gets real. Jason’s focus has shifted toward applying AI expertise in specific communities and industries, including running and outdoor.
Roast My Strava as a case study in “one-person software”
Roast My Strava came from a simple mashup:
People were prompting ChatGPT to roast their Instagram feeds.
Jason had prior experience with the Strava API.
So he built a lightweight workflow:
Connect to Strava
Pull profile + recent activity context
Run it through a large language model with a “roast” prompt
Ship it, post it, let the internet decide
He posted it on Reddit anonymously, and it caught. The most important metric for him wasn’t conversion rate or time-on-site. It was: do people want to share it?
Jason’s framing here is useful for brands: “What would it look like if your favorite meme account built software?” Code becomes the creative medium. Products can be “drops,” not multi-quarter roadmaps.
The three-phase AI implementation framework
Jason explains AI adoption in three phases. The order matters.
Phase 1: Experiment with “childlike wonder”
Try tools without fear. Use AI instead of Google for research. Ask it to rewrite something five different ways when you know what you mean but can’t land the sentence.
The point is reps. You’re building intuition.
Phase 2: Apply AI systematically to real work
Move from “one-off” prompts to repeatable workflows.
Look for tasks that are a mix of:
repetitive
but still require judgment or creative polish
Examples Jason mentions:
recording meetings and turning them into notes
scheduling automation
reusable prompts for common email patterns
The goal: buy back time for the parts of your work that actually need a human.
Phase 3: Build custom tools
Once you understand your workflows, you can graduate to:
custom GPTs (instructions + knowledge base)
lightweight internal tools
AI-assisted systems that draft content and responses for human review
Jason shared a customer support system he built that learned from his past handwritten replies + company documentation to draft future responses, which humans then review and send.
Practical implementation for outdoor brands
If you’re on a marketing team at an outdoor brand, Jason’s advice starts simple: identify the tasks you dread. You’re likely writing a ton of content and copy, some of it monotony disguised as creativity.
Use AI so you can spend more time on:
creator relationships
customer conversations
campaign strategy and creative ideation
Two “unlock” ideas he emphasized:
Give AI deep context
The quality jump often isn’t about fancy prompts. It’s about context.
When AI understands your brand, audience, products, constraints, and current projects, it becomes a reusable partner instead of a blank-page toy.
Voice-to-text is underrated
Jason swears by voice as a way to get more context into the system faster. You can speak far more nuance than you can type, which tends to produce better outputs.
He mentioned using Whisper Flow for dictation and prompting, and also using ChatGPT voice mode to literally talk through problems while walking.
Creating digital experiences people actually share
This was one of Jason’s spiciest claims: every brand can be a software company now (or at least behave like one).
The marketing question shifts from “What content should we publish?” to:
Can we build a digital experience attached to the brand that people genuinely want to share?
Examples discussed:
The Feed-style “stack” pages that show an athlete’s product setup (something anyone could use, not just sponsored athletes)
Race experiences that generate shareable graphics with personalized stats after finishing
Tools that help people prepare for activities (gear checklists, pacing calculators, training prompts, trip planners, etc.)
The barrier to prototyping is way lower now with tools like Claude and other “build from natural language” platforms. You don’t need to ship a perfect product. You need to ship something interesting enough that people want to pass it along.
Angel investing in the outdoor space
Jason’s take: angels often matter more than big venture funds, especially early.
Why?
Distribution: strategic angels often have the exact audience you need
Tactical expertise: founders one or two steps ahead give the most useful advice in crisis moments
Specialist bench: a diverse set of angels can help you vet hires, partnerships, and decisions in areas where you’re not strong
On “how to find deals,” he’s blunt: be visible and be active. Write, share your experience, and do cold outreach. Many of his best relationships (and investments) started with someone simply reaching out.
And on “how to decide,” he agrees it’s often vibes-based at the earliest stage because the company will likely pivot. You’re mostly betting on the founder, their judgment, and their ability to navigate the hard moments.
The shift toward one-person software companies
When building gets cheaper, small teams can do what used to take dozens of people and millions of dollars. Jason’s story is basically a case study in that reality.
For outdoor brands, it means:
you can run more experiments without massive commitments
you can build interactive experiences instead of only buying attention
you can operationalize AI beyond “we tried ChatGPT once”
The playbook is simple:
experiment with curiosity
apply it to repetitive creative work
build custom tools once you understand your workflows
Top takeaways
AI implementation follows three phases: experiment, systematize, then build custom tools.
One-person software is real now: the cost to build and ship digital experiences has collapsed.
“Shareability” is the metric: if people want to share the experience, you’re onto something.
Voice-to-text is a cheat code: you can feed better context faster, which improves outputs.
Angels can beat VCs early: distribution + tactical founder experience often matters more than brand-name capital.
Listen to (and watch) the episode
Stay connected
Connect with Jason Kuperberg on LinkedIn
Check out Roast My Strava and HyperWrite
The Roast My Strava case is a perfct example of how buiding for shareability beats building for sophistication. I've been tinkering with AI tools for about a year and the progression from "wow this is cool" to actually systematizing workflows really does take that childlike experimentation phase. What's underrated here is the voice-to-text point - I started dictating rough ideas while running and it completely changed how much context I could feed into prompts. The barrier now isn't technical capability, it's figuring out what's actually worth building that people will pass around.
One related question: fitness/ health data is so unique. I have had lots of data privacy concerns in my work m. How do you see the data sharing/ protection among Strava and beyond?