I build systems that keep working — not tasks that repeat.
I started an education company and sold it. Now I run a studio that builds automation. Everything below is a real, working system: it collects information, uses AI to make sense of it, and then does something useful with it — again and again, without me. Each one has an ID number from my own build list.
Apify
Claude API
Sheets / SQLite
Messages / Docs
Telegram
Build registry
selected systems · 2024–2026Every system I build gets an ID number, so I can use it again in another company later. These are the ones worth showing.
An AI notebook for studying Indian companies
It reads company reports and earnings calls, pulls out the same 27 facts from each one, and files them in my notes app. It also marks how much I should trust each source, and it can carry on from where it stopped if something breaks.
Spots shares that suddenly get busy
Every day it checks Indian shares and looks for any that are being bought and sold far more than normal. It keeps the history in a small database and sends me a ranked list on Telegram.
Finds the right online groups
It looks through groups on Reddit and Skool and notes how big each one is, how busy it is, and how strict the rules are. So I know where my customers already spend time before I message anyone.
Writes comments for me to approve
It finds posts worth replying to — by keyword, by person, or by group — and writes a draft comment for each one. I read them and approve before anything is posted, so it never looks like spam.
Posts to every social site from one place
All my posts live in one sheet. The system rewrites each post to suit each social network, schedules it, and keeps a record of how well it did.
Turns outreach numbers into a dashboard
It takes the numbers out of my LinkedIn outreach tools, tidies them up, and shows them on a live dashboard that I check once a week to decide what to do next.
Turns sales calls into records
When someone books a call, the system writes down everything that was said, picks out the parts that matter — who they are, what they want, how big the deal is — and saves it all in one place. I never take notes and nothing gets forgotten.
Matches new companies with the right investors
It compares a young company against a long list of investors — what they like to fund, how much money they put in, how early they invest — and gives back a ranked list of who to talk to first.
Finds jobs and writes the application
It collects job posts from lots of sites, adds details about the company and the person hiring, then writes a CV, a cover letter, and a plan for that exact job — all joined into one PDF, ready to send.
Finds products worth selling on Amazon
It collects product listings from Amazon India and gives each one a score: how many people want it, how many sellers there already are, and how much money would be left over. The best ones get sent to me on Telegram.
Builds a personal offer for each customer
It matches what is in stock to what each customer likes, then puts together an offer just for them, on its own. Built for expensive things that people think about for a long time before buying.
Companies
where these systems runScale Premier
A studio that helps build companies, sometimes as a part-time co-founder. I use the same set of tools every time — collecting data, scoring it with AI, outreach, CRM, reporting — and a small team that owns a share of what we build.
Company buildingLearneca
A learning platform for companies and working people, built to feel like a game. Live classes, real projects, and learning by doing. I built it, grew it, and sold it.
Education · SoldFundraising Advisory
I run the systems behind a firm that helps Indian startups raise money: matching them to the right investors, sending the outreach, and keeping track of every deal.
Startup fundingHow I work
how I choose what to buildEvery project leaves something behind
I don't do one-time jobs. If a project doesn't leave behind a tool I can use again, I change the plan or I say no.
A person always checks first
The system writes and sorts. A person says yes or no before anything goes out. Writing a draft and then approving it beats letting the machine run alone — every time.
Get the boring parts right first
Clear checklists, clear names, backups, and a way to recover when something breaks. Doing this early is the reason a system keeps working instead of slowly falling apart.
A small team that does a lot
The tools are built so 4 or 5 people can do the work of 50. Hiring more people is my last option, not my first one.
Weekly log
every Monday · what I learnedOne short note a week: what I built, what broke, and what I would do differently. Newest first.
Want someone who builds real systems, not slide decks?
I'm open to jobs and projects in automation, AI, and building companies. Happy to show you any system above, live on a call.