No updates on Kotora just yet. Baby is due any day now and I’m on week #3 at my new role. Startup vibe so I am enjoying myself. Maybe I can finesse a relationship with the CEO for funding 🤔… I’ll let you guys know how that goes.

Update

Progress has slowed a bit since my kids came back, as I expected it would, so this will be a shorter update. I haven't had as much time to work on Kotora. Busy helping a very pregnant wife with our other two kids. Baby #3 should be coming some time in February and I am not ready.

Some of last month was spent experimenting with smaller Mistral models for speaker detection and other features. Learned about Vast.AI's new serverless GPU cloud so I'll probably be going with that for the GPU infra.

A couple of new TTS models and papers have released so I've been slowly reading up on new tech so that when I have more time and money I can adapt to what is edge tech quickly.

Future updates may be nothing-burgers for the next couple of months after baby #3 but I will make an effort to finish what I have started with Kotora. Currently have too much risk around the project that I need to deal with first before continuing.

I've accepted an offer as a full-time Sr. SWE so at least my family will have health insurance and I'll have a steady income. That should de-risk things a bit but will only leave me with evenings and weekends. I think I've made enough progress that this should hopefully be enough time. My mother is also flying in to help with the baby so I will be taking full advantage.

I am also going to start doing consulting on the side and launching a couple of other SaaS businesses that are less technically challenging to get other streams of income. The idea being that if I can get these easier side-hustles to produce, I can replace my full-time income and then quit and get back to work on Kotora.

It's a temporary pivot to secure funding and make sure that moving forward I can still take big bets.

I thought I'd be further along. Actually I thought I'd launch two months ago, typical first-time founder delulu. It's hard to estimate how long a journey will take or what you will need along the way if you are traveling to a place you've never been to.

Update

Hey there, sorry for the delay on this update. I've been balancing founder life and family since my kids got back from spending time with their grandparents. Respect to solo founders building with kids.

In my last post I mentioned working on CI/CD after getting tired of breaking prod.

My goal was a clean dev flow:

• Create a task and branch
• Develop locally with HMR
• Open a PR
• Run quality checks
• Deploy the PR to staging
• Verify, merge, auto-deploy to prod, and clean up staging

Cloudflare does not really support environments. You have to create resources per env yourself. One small misconfiguration can cause staging to call prod APIs. Local dev tooling is also rough, especially with Nuxt and Nitro.

So did I get it working? Yes.

Refactoring

I refactored the project into a proper monorepo using pnpm workspaces. This caused some local dev friction, but those features were not well supported anyway.

Development

This exposed major limitations in Nuxt and Nitro. Cloudflare supports local dev with Miniflare, but Nitro does not use it. To fix this, I built a custom Nitro preset for dev that replaces cloudflare_module. It supports external Worker projects, Durable Objects, and HMR. I plan to open a PR to NitroJS so others can benefit from this.

Workflows

Using GitHub Actions, I set up workflows that detect which sub-projects changed and only run checks on those. Each PR gets its own isolated env by prefixing Cloudflare resources with the PR number and generating Wrangler configs on the fly. Staging URLs are also PR-specific and easy to share.

Conclusion

This took weeks but development is now smooth and much faster, and I am no longer breaking prod! I plan to release an example repo or get the Nitro PR merged so everyone can use this setup.

What's Next?

I have been rethinking pricing and ICPs. Early on I priced too low due to non-ICP feedback. My next update will focus on pricing and value.

Until next time!