Most outsourcing fails on visibility, not talent
When an offshore engagement goes wrong, the post-mortem almost never says the engineer could not code. It says nobody knew what was happening until it was too late to correct — progress was reported rather than observed, decisions were made in a channel the client was not in, and the invoice arrived as a single line item.
That is a solvable problem, and it is solvable with software rather than with promises. So we built the software.
The client portal described below is not a brochure feature. It is a real application, maintained by the same engineers who do client work, and it is the single largest piece of internal investment we make.
What we built
A client portal, so you can check instead of ask
Every OffSetup client gets an account. It exists so that the answer to “where are we on this?” is a page you can open, not a message you have to send and wait on.
Project tracking, step by step
Every project is broken into steps with its own status, so you can see what is done, what is in progress and what is queued — without asking anyone for an update.
Shared files where the work is
Documents, designs and deliverables attach to the project they belong to, rather than living in an email thread somebody has to go find.
Threaded discussion on the record
Comments sit against the project and against individual steps, so decisions are recorded next to the work they affected. Author identity is stamped server-side, which means a reply cannot be forged.
Billing you can audit yourself
A full statement of charges and payments, with downloadable invoice and receipt PDFs. If you want to know what you were billed for in March, you do not have to email us.
Access scoped by role
Your company's data is visible to your company and to nobody else's, enforced on the server rather than by hiding things in the interface. Commercially sensitive internal fields are never sent to a client browser at all.
Where we are investing
What we are deliberately getting better at
We would rather be genuinely good at a few things than available for everything. These are the areas we are putting real practice behind, each of them because we have already done the work and want to do more of it.
Deepening
Legacy modernisation
We keep meeting companies with a working application in an ageing stack who have been told their only option is a full rewrite. We think that advice is usually wrong, and we have built real practice around the alternative — treating the legacy system as an API and migrating incrementally, route by route, with no downtime. It is a slower story to tell and a much better outcome.
See how we did itDeepening
AI in production, with a human in the loop
We have shipped AI into a live consumer product — a classifier working from product imagery across a daily pipeline of more than thirty retailers. The part we care about is the discipline around it: low-confidence results route to a person instead of being published on trust. We are investing in that pattern, because the failure mode of an unsupervised model is not that it stops working, it is that it keeps working while being wrong.
See how we did itDeepening
Multi-tenant SaaS architecture
We run two of our own products on a database-per-tenant model — a payroll platform and a building management platform — where each customer gets a physically separate database rather than a shared table with a tenant column. It is more work to operate, and it converts the most dangerous class of bug in multi-tenant software from a forgotten query filter into something you would have to do deliberately. Payroll is not a good place to find out your isolation was only a convention.
How the isolation worksOngoing
Client-facing transparency tooling
The portal below is our own product, built and maintained by the same team that does client work, and it is where most of our internal engineering investment currently goes. Every capability we add to it exists to remove a question a client would otherwise have to ask us by email.
What is in it todayOur standards
Four things we won't do
We don't put anyone forward we haven't assessed
Every engineer on the bench has passed a technical assessment in their stack, completed a practical coding challenge and sat a culture-fit interview with our team. We shortlist two or three people and you run your own final interview. We have never pushed a placement, and if the fit is wrong we replace them at no cost.
We don't name our clients to win work
A lot of our work is embedded on someone else's product and much of it is under confidentiality. We write about the engineering in detail and leave the client anonymous. References happen directly, with permission — not on a logo wall.
We don't recommend a rewrite by default
A rewrite is the most profitable thing an agency can sell and the most reliable way to damage a business that is currently trading. When a system works, our starting position is to modernise around it rather than replace it.
We don't run a bedroom operation
Engineers work standard business hours from a managed office in Islamabad with dedicated connectivity and on-site project managers, typically overlapping four to six hours with US Eastern or UK time. HR, payroll and tax compliance are handled entirely on our side.
Want to see the portal for yourself?
We will walk you through it on a call, with a real project in it — not a slide deck.