Quicko is a tax company. We help individuals and businesses plan, prepare and file their taxes. Through Sandbox we give developers a tax API stack, and through Quicko Pro we help CAs run their practices online. It may seem odd, then, that we also build coupon managers, URL shorteners and data modelers. That work lives in Labs at Quicko, and this post explains why we do it.
TL;DR: we would rather own our tools than rent them. And once we own them, we would rather share them than keep them to ourselves.
The problems every startup shares
Every startup sets out to solve one problem for its customers. Within a year, it is solving dozens of others that have nothing to do with it. Marketing wants coupon codes for a launch, then short links to track a campaign, then an affiliate program to grow it. The blog needs a CMS. The marketing team needs Figma files turned into emails that actually render. And lately, everyone needs a place to keep their AI skills.
None of this is what customers pay for, and none of it is unique. A tax platform and a food delivery app need roughly the same coupon manager. Most startups do the sensible thing and stitch together software built by others.
We do too. We use closed source tools like Google, Notion, GitHub, Claude, Ahrefs, Razorpay and Zoho. We also self-host open source software like Listmonk, Grafana, Jenkins, OneUptime and OpenSEO. We pick the open source option whenever we can, and every year we give time and money back to the projects we rely on.
Why we build
Sometimes the tool we need does not exist. More often it exists as a SaaS product that charges per seat, does ten things when we need two, and wants our data on its servers. For a long time, the answer for a team our size was to live with the gap. Building and maintaining internal software was something only large engineering teams could afford.
AI changed that for us. We are a small team of 15 to 20 engineers, backed by a small seed round from Rainmatter by Zerodha. With AI coding tools, an experienced engineer can now take a well-defined problem from idea to a working MVP in days. That made it practical to build the tools we kept wishing for.
But cheap code is not the same as good software. Zerodha’s CTO, Kailash Nadh, makes this point in Code is cheap. Show me the talk. When writing code costs almost nothing, the value moves to thinking, judgment and accountability. In his words, “For the first time ever, good talk is exponentially more valuable than good code.” We have always believed something similar. Our products are a reflection of the conversations we have as a team. When those conversations are shallow, the products turn out bland. When they are rich, the products turn out exciting.
AI helps us write the code, but our engineers decide what is worth building, steer and review the work, and keep maintaining it after it ships.
Once we had built them, keeping them to ourselves made little sense. These are not tax problems. They are startup problems. We release them as free, open source software that anyone can host. We want to build for the internet as much as we source from it, and Labs is one way we give back.
Helping others helps us
Building for others turned out to be one of the best things we have done for our own products. When a tool has to work for people we will never meet, we cannot lean on shortcuts or tribal knowledge. It has to stand on its own.
We hold Labs to the same engineering standards as our products: SOLID design, clear interfaces, documentation, versioning and backward compatibility, OAuth, role-based access control and, where it makes sense, CLI and MCP Servers for AI agents. Labs also gives us room to try technology we might otherwise pass on, like Bun, Hono, Better Auth, SpiceDB and OpenTelemetry. Skillset, for example, runs on Bun and Hono.
The best of what we learn goes back into Quicko, Sandbox and Quicko Pro. In short, helping others helps us build better products.
Done well, Labs makes us better at the work we do every day, and lets us give something useful back to the community we build on.
How we build
Every Labs product follows a few simple rules.
Self-hosted. You run it on your own infrastructure, so your data stays with you.
Free and open source. You can read the code, change it and run it however you like. As Kailash Nadh puts it: “free as in freedom and not free as in beer.”
Used at Quicko first. Each tool runs in our own workflows before we share it. If it does not work for us, it does not ship.
Small and simple. Each tool solves one problem well.
Good on its own merits. Being free is no excuse for being bad. A Labs tool should be worth choosing even if it cost money.
There is a trap in all of this, and AI makes it easier to fall into. When code is cheap, it is tempting to build everything yourself: a coupon manager this week, a CRM next week, a help desk after that. Every tool you build is a tool you have to maintain.
Our rule is simple. If free and open source software built the right way already exists, we use it and contribute to it instead of building our own. That is why we run Listmonk, OneUptime, OpenSEO and Piwi rather than writing our own versions. Labs is for the gaps, where nothing good exists yet.
What we build
Here is what Labs has shipped so far, grouped by who it is for.
For startups
Qpon: self-hosted coupon management software.
Cliq: self-hosted software to run your own affiliate marketing program.
Silo: an open source, agent-native, headless CMS.
Skillset: a self-hosted registry for Agent Skills and plugins.
For developers
Dev Tools: a Swiss Army knife of everyday developer utilities.
Sheet Tools: a schema builder and data visualizer for sheet JSON.
UMLtimate: an object modeler that generates code in popular languages.
Dyno: a data modeler for DynamoDB with code generation.
For designers
Email HTML Generator: a Figma plugin that turns designs into email HTML that renders in Gmail, Outlook, iOS and more.
UI Skeleton Generator: a Figma plugin that converts UI designs into skeleton loading screens.
For marketers
Linq: a self-hosted URL shortener that routes visitors with rules and tracks every click.
Social Media Audience Workbench: an offline audience builder for Meta, Google, X and LinkedIn Ads.
What comes next
We believe the era of growth at all costs is over. The products that last are the ones built with care, by teams willing to be patient. That is how we have built Quicko from Ahmedabad for over a decade, and it is how we intend to build Labs.
It would be easier to keep these tools to ourselves, or to skip building them and rent whatever is on the market. Rushabh Mehta, who founded Frappe, explains why a company might choose the harder path anyway: “While pragmatism buys us comforts and power, principles buy us respect.” Labs costs us time we could spend elsewhere. We think it is worth it.
Labs is not a side project. We use these tools every day, so we will keep maintaining and growing them, and keep looking for new gaps worth filling. As we all explore what AI can do, new technologies and use cases will emerge, and Labs will grow with them. Here is what we are working on now:
Skillset will support cataloging MCP servers and Agents.
Silo will add publishing workflows.
Linq is getting OpenTelemetry support.
SDK generators: we are open sourcing the code generators we use to build SDKs from OpenAPI specs.
Over the coming months, the engineers and designers who built each tool will write about it: the problem, the decisions they made and what they learned. You can already read how we built the Email HTML Generator and Skillset.
If one of them solves a problem for you, take it. Browse everything on the Labs page, run it on your own servers and tell us what breaks on GitHub. If you want to contribute, even better.
