Technologies
Technologies we use for AI, mobile and web development.
A short list, chosen on purpose. We would rather be excellent with a few tools than show forty logos. This is what our own products and most client projects run on, and how we decide.
AI
- OpenAI
- Anthropic
- Google Gemini
- Llama, Mistral, Qwen (privately hosted)
- Vector search
- Automated test sets
We choose the model by testing on your data, never by default.
AI development services →Mobile
- Flutter
- React Native
- Swift
- Kotlin
Cross-platform by default; native when the feature demands it.
Mobile app development →Web
- Next.js
- React
- TypeScript
- Node
- Python
- Laravel
Next.js first for anything that has to rank in Google or load fast.
Web and SaaS development →Data
- PostgreSQL
- MySQL
- MongoDB
- Redis
Relational unless there is a reason not to be.
The platforms we build on it →Commerce
- Shopify
- WooCommerce
- Headless storefronts
Shopify for speed to market; headless when the brand needs more.
E-commerce development →How we choose
Four rules that decide what gets on this page.
A technology choice is a ten-year decision wearing a two-week deadline, so we make it by rule rather than by mood.
- Boring wins, except in AIFor infrastructure we pick tools that have been boring for years, because boring means documented, hireable and still maintained. The one place we move fast is models, where six months old is old.
- You can hire for it after usEvery tool here has a deep talent pool in India. A stack only we can maintain would be a switching cost dressed up as engineering, and we will not build you one.
- We run it ourselves firstEverything listed runs at least one of our four products in production. Nothing gets recommended to a client on the strength of a conference talk.
- Fewer, deeperSix groups instead of forty logos. Depth in a few tools beats familiarity with many, and it is why our estimates hold.
What we deliberately avoid
Three things missing from this page on purpose.
The framework of the month. New tools earn a place here after they have survived a year of other people's production incidents, not before. The exception is AI models, where waiting a year means shipping last year's quality.
No-code platforms for products meant to scale. They are excellent for testing demand in a fortnight, and expensive to escape once real customers and real edge cases arrive. We say this to clients even when the no-code route would be a quicker sale.
Anything that holds your data hostage. Your code, your accounts, your database exports, from day one. A stack you cannot leave is not a stack, it is a lease.
The clearest proof of the whole page is the site you are reading: it runs this exact stack, and we published the rebuild case study with the before and after scores so you can check it.
Next step
Tell us what you want to build. We will tell you what it costs and how long it takes.
A free 30-minute call with an engineer, not a salesperson. You leave with a clear plan, a price range and an honest opinion on whether AI is the right tool for the job.