Google Cloud Platform, GitLab, Slack, Close.io, Stripe.<p>Here's the stack:<p><pre><code> Stack:
- Flask
- PostgreSQL
- Monitoring:
- Prometheus
- Grafana
- Sentry
- Analytics:
- PostHog
- RudderStack
- Containers:
- Docker images on GitLab registry
- Repository management:
- GitLab:
- Issue templates (lower the barrier to write good issues with templates):
- Bugs
- Features
- Incidents
- Communications:
- Slack
</code></pre>
I would tackle distribution way earlier. We are privileged in the product we're making in that it started as an internal machine learning platform for us (we were extremely familiar with the problem space), because we're a tiny, boutique, machine learning consultancy, and even being in that position, we treated the internal platform as if it were a public facing product from the get go: we put it live, had a landing page, onboarded users on a request basis, reached out to to people in the field to have conversations to learn what problems they're having and to solve a more generalized version of the problems we were solving for ourselves, and to bring focus to our backlog by talking with potential users who did not want to use it, and figuring out why.<p>Notwithstanding that, I would try and get it in the hands of users earlier than I would, even if it were internal.
Depends on the company, I think the only SaaS you really need is the payment provider (eg. Stripe/Paddle). Apart from that you might need hosting, I would personally go with a cheap VPS instead of cloud services.
I wouldn't. I'd use a bunch of secondhand computers running GNU/Linux or BSD until I've gotten off the ground and actually need to dick around with other people's computers^W^W^Wthe cloud.