How Rara Labs Put Claude to Work Across the Whole Company

Rara Labs is a product design and development company based in Nepal, with more than 100 people building software for clients across fintech, ERP, healthcare, and AI. It's the kind of company that lives on range: a ledger engine for one week, a hospital scheduling tool the next, an internal AI product after that. Today, Claude Code and the Claude API run through engineering, product design, and prototyping company-wide, and the team ships more without adding headcount to do it.

The Challenge

At any given time, Rara Labs runs more than ten projects at once, spread across industries that don't share much besides the need to get it right: a financial ledger can't drift by a cent, a healthcare tool can't cut corners on reliability. Every project asks something slightly different from the people building it — new domain knowledge, an unfamiliar codebase, its own conventions and constraints.

That range creates a staffing problem with no clean solution. Hiring a specialist for every domain doesn't scale for a company of this size, and spreading a lean team across everything means something must give. In practice, what gave way was the unglamorous but necessary work: the refactor that would keep a system healthy for another two years, the cleanup that made a codebase easier for the next engineer to enter. It wasn't that Rara Labs lacked ideas for what to build or fix — it was that every idea had a real cost in engineering hours, and the queue for those hours was already full.

What They Built

There were no top-down mandates and no company-wide launch day. Claude moved into Rara Labs the way a genuinely useful tool usually does: one team picked it up, found it solved a real problem, and word spread from there. It took hold first in engineering, then followed the work into product design.

Engineering is where the habit runs deepest. Rara's developers now reach for Claude Code and the Claude API across nearly everything the job involves — building features end to end, debugging (tracing a problem back to where it started, not just patching a symptom), and writing and running the tests that keep new work from quietly breaking old work. Because the team rotates between unrelated codebases in the same week, they lean on Claude heavily just to get oriented in an unfamiliar system before touching it.

That same fluency made room for deferred work. Engineers now use Claude to optimize and refactor existing systems — reshaping code built for an earlier, smaller version of the product so it can hold up at the next one, instead of accepting a slow, risky rewrite as the only option.

Product design and prototyping picked it up next. Designers use Claude to take a rough idea — often little more than a sentence — and turn it into a structured spec the engineering team can build from. They use it to explore and iterate faster than a single designer could alone, and instead of stopping at a static mockup, they use Claude to stand up a working prototype so a stakeholder can click through something real. The specs and designs that come out of that process are documented with Claude too.

• Engineering: Claude Code and the Claude API across the full workflow — building features, debugging, testing, understanding unfamiliar codebases, and refactoring for scale
• Product design: from a rough idea to a structured spec to a working demo, with the resulting designs documented along the way

The Results

Ask Rara Labs what changed, and the answer isn't a single metric — it's a description of what the same team can now carry. Projects that used to compete for the same few senior engineers no longer have to wait in line, because work that once required a senior review can now be finished by someone earlier in their career, with Claude filling in the parts of the job that used to require years of accumulated pattern-matching.

That has a compounding effect on the backlog itself. Ideas that would have quietly stalled — good ones, just not urgent enough to win a slot on a stretched roadmap — now get built, because the distance between having the idea and holding a working version of it has shrunk. And by Rara's account, the speed hasn't come at the cost of quality — if anything, the two have risen together.

Rara Labs now puts a number on it: a project that once needed 5–6 engineers to staff can be run with 2, without a drop in output. The team is quick to note this isn't a claim about tooling alone — it reflects Claude handling a meaningful share of the work that used to require additional dedicated staff.

“We did not set out to roll Claude across the company. It kept winning the argument, one team at a time. The engineers found it first, then it moved into how we design and prototype our products. What changed is not simply that we work faster. It is that ideas reach a working product far sooner than they used to.”

Gautam Acharya

Lead Engineer, Rara Labs