Unlock the true potential of your data.

The tech side of startups is often fluid, full of unknowns and rapid changes. Questions like which tech stack to choose, which tools are essential now vs. later, and how to balance fast feature delivery with long-term code quality are constant challenges.

In this post, I’ll share our journey building https://cleanbee.syzygy-ai.com/ from the ground up — how we tailored our processes to meet evolving needs, and how those processes evolved in turn as we expanded our tech stack and added new components.

Initiate the Process Design – formal and strategic

Where We Begin:

• A compact team of fewer than twelve, combining in-house developers and short-term contractors, aligned for asynchronous collaboration.
• This is a true greenfield initiative — no existing codebase. The requirements are still forming, but a rough structure is starting to emerge.
• On the tech side, it's evident that a backend is essential to support communication with mobile clients.
• Originally thought a few static pages would do for the web frontend — that idea didn’t last.

We started with a simple setup — code hosted on GitHub and a pull request–based workflow, guided by one key requirement: deliver a split version of the tickets table within 1–3 days. This approach encouraged the practice of story slicing, and the steady movement of tickets to 'Done' has helped create a visible sense of progress. That momentum can serve as a strong motivational driver for the team to embrace this way of working.

Our Simplified Roadmap

Today, the range of frameworks is vast, but given our experience with Flutter, we decided to stick with it. In mobile development, one of the key considerations is choosing the right state management solution.

What led us to Terraform?

The backend, including the database, will also require object/file storage. Eventually, we’ll need DNS to ensure our services are ready for the broader internet landscape.

The choice of cloud provider was based purely on the team's existing expertise. We’re already using AWS for other projects, so we decided to stay with it. While it's certainly possible to manage everything via the AWS console, over time it can become a tangled mess—what starts as a clean setup turns into a maze that no one wants to touch, and eventually, no one remembers why certain components even exist.

Our approach to selecting partners and technologies is to evaluate the most promising options based on features, supported languages, and, for providers, pricing.