Building useful software, AI products, and experiments.

RismanLabs is an independent software lab. It designs, builds, and runs focused products and practical AI tools, end to end.

Products and experiments

Each one is built as a whole thing: its own name, its own repository, and its own address under rismanlabs.com.

Featured product

DuoDue

Simple debt tracking for two people.

A focused ledger for couples who lend to each other, split costs, and pay each other back. Every entry is agreed by both of you, so the balance is a shared fact instead of a running argument.

One balance you can both see
Debts and repayments roll into a single figure that is current for both of you.
Nothing counts until you both agree
An entry joins the balance only once the other person confirms it.
Built for two, on purpose
No groups and no splitting a bill five ways. It does one relationship well.

In development · Next.js, TypeScript, PostgreSQL

Dimas owes Ayu

Shared ledger

Rp 3.090.000

  1. Flights to LombokAdded by Ayu, confirmed by both+Rp 1.840.000
  2. Half of October rentAdded by Ayu, confirmed by both+Rp 2.250.000
  3. RepaymentAdded by Dimas, confirmed by both-Rp 1.000.000

Try it. Nothing here is saved.

A working sketch of the idea, not the shipped interface.

About

RismanLabs is run by Muhamad Risman, a senior full stack engineer who builds scalable web applications and, increasingly, products with AI doing real work inside them.

12+

years shipping production software

Interface

  • Next.js
  • React
  • TypeScript

Services

  • Golang
  • Python

Intelligence

  • AI integration

How the lab builds

Four commitments every product here is held to, from the first commit to the last deploy.

Clean architecture

Boundaries are drawn around the domain, not the framework, so a product can change its mind without a rewrite.

Performance

Rendered on the server, measured before release, and held to a budget rather than tuned after complaints.

Developer experience

Typed end to end, tested at the seams, and dull to deploy. Code the next person can read on the first pass.

Practical AI

Models go where they remove real work for the person using the product, and nowhere they would only decorate it.