This is a blog post about how relying on coding agents are making software engineers who create every intricate function of a codebase, into farmers who grow codebases, doctors who make changes on these systems which they have not created and gardeners who try to the same process of farming to make the change.
I am Aswin (https://aswinmohan.me) and I am building Cycle. It's an integrated financial platform that combines business banking, an inbuilt accounting ledger and payroll. By combining them together we get additional context which we can use to automate much of the grunt-work associated with managing finance.
You can try out the accounting system here with no sign-up at https://demo.joincycle.co. You can join the waitlist to use it when we launch with business banking.
I have built the platform in Elixir, Phoenix, Inertia JS and React. I had launched it initially as workbill.co, as a standalone accounting platform, but later pivoted to this.
Unlike traditional accounting platforms we expose the ledger model directly which enables our customers to model complex transactions even when we do not have direct support for it.
Been working on this for a month, and it uses Elixir, Phoenix and InertiaJS with React.
We use a proper ledger to track the flow of money. It was sizable technical investment to get the schema right, but it has made the subsequent implementations of features easier.
You are right about missing the web part. It's substantial space that we miss out on mobile. But mobile lets you enter transactions on the go, and is always there when you need it. A web version is planned in the future.
I started Paper to bring the beancount + automated parsing with python stack to everyone. Double-entry is an objectively better way to manage your finances, when we build an automatic reconciliation engine that automatically categorizes the bank transactions we'll have the best personal finance tracker app.
Automated transaction reconciliation will be a major feature in the future versions of the app. I have yet to use an app that reliably does automatic transaction registration and the dream is to build one.
India does have a decent open-banking system, used by https://fold.money
It's a characteristic of double-entry accounting. Everything is an account and transactions move money between accounts. When you receive money from your salary in your bank account, the money is being debited from an Income account. We show the debit with the negative sign.
> Yes you can, it's called a split transaction. $1000 income: split into $700 to my savings and $300 to taxes.
Even though this would work, the money is still in your bank account, which makes your savings account balance to be reported wrong.
In India we use Aadhar, the associated phone number and a digi-locker app to verify our identity. Mostly used for governmental, financial and insurance verification. Works seamlessly most of the time.
I bought a mechanical watch after reading this blog post and it's been my daily driver since. It always brings a smile to look at it and see what fine craftsmanship and mechanical intelligence can achieve.
Does a system exist where we can train open source LLM models in this way? Maybe distribute it across a cluster of computers with credit points that can be later used for inference.
I use the Moonlander with a layout that doesn't deviate much from the standard Mac keyboard. Even though I sacrifice a little bit of single keypress magics that I can do on my Moonlander, it enables me to switch more easily to my laptop keyboard on the go.
While working on my startup as a solo-founder I used to follow a TDD approach supported mostly by Integration tests. I mostly focused on the user happy paths with limited coverage outside the Integration tests.
The goal is to have the major functionality of the platform tested automatically so we can be confident that new features don’t introduce new bugs.