Gravixar is the software I build to run a business's daily operations. AI does the busywork, and a person approves anything that spends money, changes a record, or goes out the door. You see it running before you buy it.
How I work
Every system I build follows the same rule: a person approves before anything changes. When a team needs it shaped around how they work, I build it myself. There is no agency layer between you and the person doing the work. That's me.
The systems I build for clients are live and running, and I usually run a version myself before I recommend one. There are five demo apps at demo.gravixar.com: sample data, no sign-in, nothing to install. They are built separately from any client's system, on purpose. Pick the one closest to your work, click through it, and decide if it makes sense before you ever get on a call.
I build in small steps. In week one you have something running, even if it is small. By week ten you have something that has been used in real conditions and rounded off where the corners caught. I don't disappear into a four-month build and come back with a "v1" you have never seen.
Running eight live client projects at once only works because the setup behind them is a system too, with checks that run every day and tell me when something has drifted. Here is what that system handles.
Counted on 2 November 2026
- Building blocks in the library (modules)
- 31
- Of those, used in more than one product
- 15
- Tasks that run on a schedule
- 29
Work done in July 2026
- Saved code changes (commits)
- 378
- Batches of changes merged (pull requests)
- 158
- Automated upkeep tasks, each one behind a check
- 68
When AI is part of a system, it drafts and suggests, and a person approves. The AI that drafts articles for this site does not publish them on its own. The market scan hands me a ranked list and waits. Nothing on it acts by itself. That is what I sell: AI handles the boring 80%, and a person keeps the steering wheel.
What I'm working on
An agency portal
For a digital agency, live in production
A new client goes from first inquiry to an active project in about 10 minutes. By email, that took two to three days.
A healthcare credentialing platform
Live on its own domain since June 2026
For the first time, the team can see how many insurer applications are waiting on them (125) and how many are waiting on the insurers (1,560), across 250 clients.
This site
Small in scope, and still real proof
One job runs on a schedule here: it tells Google about open jobs every day. Once a week a research routine on my own machine drafts at most one article and a market brief, and the article waits for my yes. None of it publishes on its own.
The hosted platform
Not built yet
The same building blocks, rented by the month instead of built for you. It does not exist yet, and I am not printing a date for it. Join the list.
Background
I started managing operations in 2009 and have been doing it in some form ever since: HubSpot, monday.com, ClickUp, Zoho, and most of the rest of the suite over the years, on teams that were on-site, remote, and everything in between. These days I also head operations for an Islamabad-based creative agency on top of my own work, so the patterns I sell aren't things I've read about. They're things I run.
I've watched enough teams stitch Notion, Sheets and a Slack channel into something that almost works to know that "almost" is where most of the cost lives. The gap between that and a real system is where I tend to be useful.
A few things I'm not
- Not a CRM developer, though I've built portals with CRM-shaped pieces in them.
- Not a vendor who'll tell you AI is the answer to a problem AI is bad at.
- Not a black box. Every system I ship comes with a record of what it did, the code itself, and the reasons behind each decision. I have written down 1,366 decisions and 170 infrastructure notes so far, and I only ever add to them, so nothing gets quietly rewritten later.
Outside this
Islamabad-based, originally from Jabri, a small village in the mountains where I take my four boys when I can. They're all homeschooled, so a fair bit of my non-work time goes into STEM with them. Some of that became Robonamix, a robotics and STEM project for kids that I am picking back up. The rest of the time, mostly hiking.