HouzeGuru's - A Wise Home

A little around 2 years back, I started with a professional consultancy for HouzeGuru - with an idea to build a platform for managing everything at home - like an ERP - but for your house, build specifically for families. I later joined Prashant as a CTO co-founder for HouzeGuru and have ever since being building this - currently user facing in the form of an app - A Wise Home  (download for iOS and Android from here: https://awisehome.onelink.me/3xG5/custref) 

When we started of more than 2 year ago GitHub copilot barely did code completion. There there way Cody that was better imo for code blacks and auto documentation. Then I stumbled upon kiro - that completely changed my perspective on how to build features or entire products. I was an early user for kiro so that gave me fair amount of credits to pay with - which primarily I used to built my pet project - the kosh OS. Later I moved to a paid subscription and haven’t looked back - I don’t since use Kiro - use more of Claude code and Antigravity - sometimes Codex. But the entire perspective on how I write software form ideation to product has changed for me. There is bare minimum hand costing - but I still do manual code reviews even when I employ agents to do first level of coding. 

More important than not - these tools have enabled me to do things which would otherwise take ages for me to actually do due to the sheer labour involved - it is not that I don’t understand how to build but rather just the amount of keystrokes it needs to get is done was always a bottleneck for me. 

These tools have also given me hope than I would probably one day revive metastudio that I literally hand coded and designed for over a decade (c.f.https://github.com/tovganesh/metastudio ) and then just couldn’t devote time as I got involved in my other professional engagements. 

In the meantime do try out Houzeguru - which is now rebranded as A Wise Home (https://www.awisehome.app) - it is built be real humans and folks those have family and plan to have one. See how it works out for yours and let me or Prashant know!

kosh - the OS - current status

 I originally wrote down the thoughts on what I would like to have from an OS that I would like to use on a daily basis here:  https://blog.tovganesh.in/2012/01/kosh-building-mobile-user-experience.html 

That was about 14 years ago. But I never wrote a line of code for this. For two things: I simply didn't have the bandwidth do write low-level codes, and I didn't want to write this in C. Back then, I had not heard of Rust. And back then there were no agents around that could churn out code in a matter of time that have far surpassed best of humans - couple of LLM generations back. 

I got particularly exited because of Kiro - which I had been using it for some personal tryouts. I used it to write and refine first formal specs for Kosh using my blog as the basis - around Aug 2025 (https://github.com/tovganesh/kosh). But since then I have shifted to using Claude code and Antigravity. With both these I have now kind of a working kernel which I can boot into a Virtual machine. I would have never imagined getting my initial ideas on the OS experience even this far. The full experience is build around Rust - which I feel is a solid systems programming language. 

Kiro's abilities is do spec based development and its ability to make solidly testable code - at a pace that was simply not possible doing manually - is what helped me completely pivot away from manual coding. Since then I have used other coding agents for vast majority of my personal and professional projects with restricting to manual review and code for only real edge cases - or the once where it is just easier to fix manually rather then prompting to actually get things done. While I use these tools quite heavily to produce real usable features - I very rarely hit token limits. I am not very sure how other people do so frequently hit these limits, may be it is just me. But I have also found that overusing these tools can result into projects that are functional - but inside it is a complete bloat of unmaintainable code. One such project is what I am currently evaluating and figuring out how best to salvage the internals - and boy that is going to be some fun. 

 

 

Chromebooks and Googlebooks

Recently Google released Googlebook - built by many of its partners. To me there are two major concerns for this: Whom this is for? I feel that it is trying to find user for this - who doesn't exist. The pricing of this is such that it squarely falls under MacBook Air category - but then it cannot do all the things that the MacBook can do - like the true desktop things. And then there is connectivity issue - it missed a chance to not support un device cellular connectivity - something that I truly hope that the next MacBook will have. 

That brings me back to google's other product : the chrombook (the device I am currently using to write this). What does it really offer over Chromebook - higher spec and higher cost - but functionality ? Barely anything worth paying the extra premium. 

Essentially - Googlebook just doesn't compute.