May 2026
Your business probably doesn't need AI. Here's how to tell.
We make money building AI systems, which is exactly why you should hear this from us: most of the "AI ideas" that reach our inbox are spreadsheets in a trench coat.
Keep reading
Here's the test we use in-house. First: is the task repetitive, high-volume, and tolerant of the occasional mistake? AI thrives there. Second: would a clear form, a filter, or one good database query solve it? Then build that instead; it's cheaper and it never hallucinates.
One client came to us wanting an AI to "predict inventory." After two days on-site we found the real problem: three branches counted stock in three different units. We fixed the counting. No model required. They saved the AI budget and the embarrassment.
When AI does fit, as it did for a helpdesk we automated, the wins are real and measurable. The trick is letting the problem choose the tool, never the other way around.
March 2026
The seventy-minute waiting room, and what it taught us about design
Before designing a clinic app, we sat in the waiting room for two days. Not interviewing anyone. Just waiting, like everyone else. It changed the whole product.
Keep reading
From a desk, the brief looked simple: "patients want shorter waits." From a plastic chair at hour one, the truth was different. People didn't mind waiting nearly as much as they minded not knowing. The sighing started when the queue felt random, when someone who arrived later went in first, when nobody could say "you're third."
So the app we built barely tries to shorten the wait. It makes the wait honest: your live position, a realistic estimate, a nudge when it's time to leave home. Average time spent physically waiting dropped by half, not because the doctor got faster, but because people arrived when it mattered.
The lesson we keep relearning: the problem the brief describes is rarely the problem the chair reveals. Go sit in the chair.
January 2026
Why we still sketch on paper in 2026
Our design tools are excellent. Our pencils are better. A defence of the humble paper sketch, from a studio that ships modern software for a living.
Keep reading
A polished mockup is dangerously persuasive. Show a client a pixel-perfect screen and they'll critique the shade of teal; show them a pencil sketch and they'll critique the idea. That's the feedback that actually matters in week one.
Paper is also gloriously cheap to be wrong on. We've watched teams defend a bad layout for weeks because someone spent two days polishing it. Nobody defends a sketch. You cross it out, you draw the next one, your ego stays out of the building.
So every AKARA project still starts with printer paper and a pencil cup, and the wall of crossed-out ideas is, honestly, our favourite decoration in the studio.
November 2025
The slow website is costing you more than the fast one would
We tested a client's old site on a five-year-old phone with patchy 4G, the device their actual customers use. It took eleven seconds to load. Here's what that means in lost rupees.
Keep reading
Eleven seconds doesn't sound catastrophic until you watch someone live through it. By second four they've checked their notifications. By second seven they've mentally filed your business under "maybe later." By second eleven, the page has loaded and they're gone.
The fixes are rarely glamorous. We compressed images that were ten times larger than needed, removed four tracking scripts nobody had looked at in a year, and let the text load before the decorations. Load time fell to under two seconds on the same battered phone.
Enquiries from the website went up by a third the following month. Same business, same prices, same design. The only thing that changed was respect for the visitor's time.
September 2025
What "maintenance" actually means, in plain language
Software maintenance sounds like a fee invented to pad invoices, until the day it isn't. Here's what's genuinely inside that line item, explained like you'd explain it to a friend.
Keep reading
Software lives in a world that keeps moving. Phones update, browsers change, the payment gateway revises its rules, a security hole gets discovered in a library half the internet uses. None of this is your app's fault, and all of it is your app's problem.
Maintenance is the quiet work of absorbing those changes before your customers feel them: updates applied, backups tested (not just taken, tested), small bugs fixed while they're still small. Done well, it's invisible. Skipped, it's a 2 a.m. phone call and a very visible outage.
Our rule for clients: if a maintenance report ever feels boring, that's the product working. Boring is what you're paying for.