2023: Shutting Down a Hit App, AI Goes Mainstream

2023 was the year Strato Foods, my biggest Strato Inc app, stopped making sense to operate. It was also the year AI went from “interesting research area” to “every dinner conversation mentions ChatGPT.” I shut down Strato Foods in February, kept publishing through Strato Inc, and spent the rest of the year learning what building looks like after your local hit dies.

This is not a highlight reel. It is an honest accounting.

The Year at a Glance

timeline
    title 2023 - Personal and Professional
    section Q1
        Jan : GPT-4 launch reshapes AI discourse
        Feb : Strato Foods shutdown under Strato Inc
        Mar : SVB collapse triggers treasury lessons
    section Q2
        Apr-May : Recovery and reflection post-shutdown
        Jun-Jul : Small Strato Inc releases, summer building
    section Q3
        Aug : More Strato Inc app experiments
        Sep : Vision transformer production reading
        Oct : Continual learning research deepens
    section Q4
        Nov : OpenAI DevDay and GPT Store hype
        Nov : OpenAI board drama governance lessons
        Dec : JEE prep and year-end consolidation

January-March: AI Hype and Hard Lessons

GPT-4 dropped in March and reset every conversation about what machines can do. I had been working with LLMs since GPT-3, but GPT-4 was a step change that made non-technical people suddenly care about context windows and hallucination rates.

I shut down Strato Foods in February. I wrote about that separately. The meta-lesson: product success in a local market is not the same as a durable business when national aggregators arrive. Strato Inc continued. The app did not. That distinction mattered for my sanity.

SVB collapsed in March. Even as a mostly India-based builder, the panic was a reminder that banking and treasury hygiene are founder skills, not finance team problems.

April-September: Research and Strato Inc

Q2 and Q3 were about shipping smaller things through Strato Inc and reading ML papers without pretending each paper was a company pitch. Class 12 and JEE prep were the other calendar — sometimes winning, sometimes not.

On the research side, I spent time on continual learning for LLMs, vision transformers, and what production actually needs versus what papers claim.

October-December: Consolidation

Q4 was writing, research consolidation, and planning for college. OpenAI DevDay in November was a builder-economics moment. The board drama the same month was a governance lesson for anyone building on top of foundation models.

Numbers I Will Share

What 2024 Needed to Be

Keep Strato Inc alive. Finish school. Get into college without abandoning the builder identity that survived the shutdown.

Closing

2023 was the year I learned that shutting down Strato Foods is not shutting down as a builder. Strato Inc outlived the app. The skills compounded. The identity crisis was real and temporary.

Happy new year. Back to work.

--claps