Self-use proof
The system my own studio runs on
a normal day in it, part by part
Before I sell a system I run it. My studio's attendance, leave, tasks, projects and CRM are all mine, and nobody there sat through a training session to use them.
I built this for my own studio before I sold a line of it. Eighteen people in Mumbai use it every day. Nobody was trained on it. This page walks through a normal day and says what each part replaced.
Cove Craft Studios is a 3D rendering and AI media studio in Mumbai. It has eighteen people, a lot of files, and the same problems as any small company: who is in today, who is on leave, what the artists finished, where that client enquiry went. None of it is answered by asking around any more. It is answered from records the studio already keeps, on a machine in the office.
Eighteen people, no training session
Nobody at the studio was taught how to use it. If a system needs a training session, I have built it in the wrong place, so I move it into the chat the team already has open.
I ask the records, I do not read them
Project status, attendance, who is free this week. I type the question into the studio chat and the answer comes back with the record it was read from.
It restarts itself after a power cut
Mumbai loses power. The machine comes back on its own, starts everything in the right order, and messages my phone if it cannot reach the internet from the office.
Nothing leaves the building
The records, the chat history and the model sit on a machine in the studio. No outside AI service holds a copy of what my team or my clients send.
Built once, with no monthly licence
There is no software subscription behind any of it. The studio hired its eighteenth person without a single line on the bill changing.
Morning: the gate, the leave board and the day's work
The first thing anyone touches is the fingerprint reader at the door. It writes straight into the attendance record, so there is no register to fill in and nobody reconciling a register against a payroll sheet at month end. If somebody is working from a shoot or a client's office, they check in from their phone instead, with a photo taken at that moment.
Leave used to be a message to me, which meant it lived in my phone and nowhere else. Now the request goes into the system, the person who approves it is asked, and the leave is on the board before the team plans around it. It is the same board the managers read for the day's work, so the two do not drift apart.
- Fingerprint attendance, written to the record as it happens
- Off-site check-in from a phone, with a photo
- Leave requested, approved and visible in one place
- The day's tasks on a board the managers already read
- No register, no email chain, no separate payroll sheet
Through the day: asking the records instead of reading them
The studio runs on a team chat, and the management Agent sits in it. An Agent is a program that understands what was asked, does the work in the apps my team already uses, and writes down what it did. I ask it which projects are behind, who has nothing assigned, what a client was last told. It answers from the studio's own records and names the record it read.
That is the part that changed how I work. I used to open four screens and assemble the picture myself, usually late. Now I ask one question in the chat I am already in. Nobody added an app to their phone for this, which is the reason it gets used.
Anything that changes a record or sends a message is confirmed by a person first. The Agent asks, I say yes, and the action goes into a log I can read afterwards.
Evening: the day writes itself up
At the end of a shift each artist sends what they did as a short message or a voice note. The system turns that into the day's record against the right project, and the managers get one summary instead of chasing eleven people for updates. Nobody writes a report for the sake of a report.
The enquiry side works the same way. A new enquiry is logged, followed up, and the follow-up is written down, so when a client calls back nobody has to remember what was promised in a chat that has scrolled away.
What keeps it standing when nobody is watching
A system that needs me is not finished. The studio machine restarts everything in the right order after a power cut, which in Mumbai is a routine event. If it cannot reach the internet it messages my phone before anyone in the office notices. If an Agent fails a job, it says so in the chat rather than failing quietly.
This is the part no demonstration shows, and it decides whether a system is still in use a year later. It is also why I run my own studio on its own hardware first. The faults showed up in my office, at my cost, before anyone paid me for the same build.
What a system like this costs to build
I quote this in parts, because almost nobody wants all of it at once. Attendance and leave is one part. The task and project record is another. The Agent in the chat is a third. Most offices start with the part costing them the most time and add the next one later.
The price follows an audit, and the quote names each part, the machine it runs on and the handover. What moves it: how many parts you want, how much of your record keeping is still on paper, how many people have to approve things, and whether a computer you already own can do the work. There is no monthly software subscription for any of it. If you want me to keep an eye on it afterwards, that is priced on its own and you can end it whenever you like.
How it runs
- 01
Audit
I sit with the people who do the work and map how the day really moves, including the parts that work fine and should be left exactly as they are.
- 02
Pick one part
We choose the part costing the most time, usually attendance or the daily work record, and agree what that record should look like at the end.
- 03
Build it on your machine
It is built on a computer you own and run with one team until it survives a normal week with nobody looking after it.
- 04
Add the next part
Once the first part is used without being asked for, the next one goes in. The machine, the code and the accounts are yours throughout.
Built, not described
Questions people ask
Can I see the system you run your own company on?
Yes. The studio's own systems are on my projects page and several of them open in a browser with no form and no call. What you cannot see is my team's live data, so the demos run on invented records that behave the same way.
How much does a system like your studio's cost?
There is no fixed price, because I quote it in parts and almost nobody wants all of them. The quote follows the audit and names each part, the machine and the handover. There is no monthly software subscription for the system itself.
Did your team need training to use it?
No. The systems live in the chat and on the phone the studio already used, so there was nothing new to open. There was nothing to train anybody on. If a build needs training, I have put it in the wrong place.
What happens when the power or the internet goes?
The machine restarts everything in order after a power cut and messages my phone if it cannot reach the internet. The model runs locally, so answers keep coming. Anything that has to go out, like a WhatsApp message, waits for the line to come back.
Does this only work for a creative studio?
No. Attendance, leave, daily work, projects and enquiries look much the same in a factory, a clinic, a hotel or a design practice. What changes is the records underneath and who approves what. The audit is where that gets decided.
Is any of the data on an outside service?
The model, the records and the chat history sit on a machine in my studio. That is the same arrangement I build for clients, on hardware bought in their name. Messages still travel over WhatsApp, which is the one place work leaves the building, and I say that plainly in the design.
Can I speak to the people who use it?
Ask me and I will arrange it. My studio is in Mumbai and the people who use these systems every day are there. I would rather you heard what annoys them about it than read a quote I chose.
Next step