Deep-tech studio shipping production ERPs, applied AI, healthtech & edtech. We build in public: the problems, the trade-offs, the wins. Plus hackathons.
Everyone's an "AI Engineer" now. But it's four different jobs:
• ML Engineer, builds the model
• AI Engineer, builds the product
• Agentic AI Engineer, makes it act
• Forward Deployed Engineer, builds it with the client (top paid)
Which one are you?
Apple confirmed Siri will run on Google's Gemini. Reported: ~$1B/year for a custom 1.2 trillion parameter model.
The richest company on earth decided not to build its own frontier AI. It rented one.
The model isn't the moat anymore. Distribution, integration, and trust are. 👇
We didn't build AarogyaOS in a room full of devs.
We sat with doctors from top hospitals and watched where software gets in the way. Every extra click pulls attention off the patient.
So we built the opposite: fast, calm, out of the way.
Here's the queue 👇
AI. ML. LLMs. Agents. MCP.
Everyone says them like they're the same thing. They're not.
A 20-second clip from our AI workshop, mapping how the whole thing fits together 👇
Which one tripped you up the most?
"Can you add AI to it?"
We get asked this weekly. We ask "to do what?" and there's a pause.
"Add AI" isn't a plan. It's fear of being left behind with a label on it.
The AI that lasts solves a problem you could've named in one sentence before AI existed.
Poll, be honest 👇
The fix:
pause animation when the tab is hidden or offscreen
serve a static fallback when there's no GPU or reduced motion
keep your text server-rendered so Google can read it either way
Our homepage looked incredible. A 3D WebGL hero, hundreds of animated particles, buttery-smooth scroll.
Then we ran PageSpeed Insights and it just… failed.
"The page stopped responding."
What we learned about pretty sites and SEO 👇
Worse: those tools run with no GPU. Your WebGL falls back to software rendering and basically hangs.
And Googlebot renders the same way. So a homepage like this can end up barely indexed.
The hero we were proud of was quietly making us invisible.
We went quiet. That was a mistake, because we never stopped building.
Over the last few years, DevelopersBracket has been heads-down shipping real software: ERPs running live in classrooms, healthcare and restaurant platforms, and AI products people actually use.
We just never talked about it. That changes now.
From this week, we're sharing how we build: the hard problems, the trade-offs, the wins, and what we'd do differently. No fluff, just the engineering.
If you build software, run operations, or are curious how modern products get made, follow along.
First story drops Tuesday. 👇
hashtag#buildinpublic hashtag#softwaredevelopment hashtag#saas hashtag#startupindia