Khadeem Smart
A student in the Netherlands who designed and built an entire business suite, its agents, and the platform that runs them in production.
Most business software is designed by people who have never had to run a business, and sold by people who have never had to use it. That gap is why a warehouse manager ends up keeping a private spreadsheet, and why a finance team spends the first week of every month working out which of two systems is lying.
Khadeem Smart started from the other end. Before writing the software he went and understood how companies actually operate: how an order really moves from a phone call to a delivery note, where the stock figure genuinely comes from, what a bookkeeper does at month end that no vendor has ever bothered to watch. He talked to the people doing the work rather than the people signing the contract.
What became obvious is that almost every problem those companies described was the same problem wearing different clothes. The data was in more than one place. Everything downstream, from the re-keying to the reconciliation to the decision made three weeks too late, followed from that single fact.
The fix was clear and enormously unattractive: build all of it, on one database, properly. Not a CRM with an accounting bolt-on. Not an ERP with a chat feature. The ledger, the warehouse, the pipeline, the helpdesk, the payroll, the shop, the website, the files, the documents and the meetings, as a single system where a customer is one record.
He did it anyway, and he did it alone. Trivena's framework, its 18 applications, its 6 department agents, its hosting platform, its provisioning, its billing and its control plane were designed and built by one person, while studying, in the Netherlands.
The agents came from the same conclusion rather than from the industry getting excited about AI. Once every record a company holds is in one database with one permission model, a genuinely useful agent stops being a research problem and becomes an obvious feature. The interesting work was not making them clever. It was making them safe: scoping each one to a department, inheriting the asking user's permissions exactly, putting approval gates in front of anything that writes, and deciding that they retain nothing at all when the conversation ends.
He is still the person who answers the email address on the contact page. Write to him and he will tell you honestly whether Trivena fits what you are doing, including when it does not.
What he actually built.
Not oversight of a team. The design and the implementation.
The framework everything stands on
The data model, the permission engine, the document and workflow layer, the background job system and the API surface. Every application in the suite is written against it, which is the reason they behave as one product rather than eighteen.
All 18 applications
From the double entry ledger and the warehouse through to the website builder and the presentation editor. Designed against one data model from the first line, so a customer, an item and an employee mean the same thing everywhere.
The 6 department agents
The scoping model, the permission inheritance, the approval gates and the decision that agents retain nothing. The hardest part was not making them capable. It was making them safe enough to put near a general ledger.
The platform that runs it
Provisioning, isolation per customer, deployment, backups, monitoring, certificates, billing and the control plane behind all of it. Every Trivena Site in production was created by infrastructure he designed.
The rules he built it by.
Every one of these closed off something easier. Together they are why the product looks the way it does, and why the pricing page has actual numbers on it.
- If a person has to re-key it, the software failed
- Say what the system does, not what it might do next quarter
- Charge for what costs money to run, and nothing else
- A limit you can see beats a promise you cannot check
- Build the boring part properly and the rest becomes easy
I did not want to sell companies another system. They already have too many. I wanted to give them one.