A project you entrust to us, or a tender we respond to. In both cases, each stage produces something you can read before the next one starts.
A project you award to us directly follows the thread below, from scoping to handover. A public or restricted tender adds the buyer's procedure upstream: reading the tender pack, questions, an optional consortium, the offer, a hearing, notification. Once the contract is awarded, we start the same scoping stage as a privately contracted project.
The second thread describes usual tender practice, not an outcome we claim to have won. Each procedure has its own rules; stages may be merged or missing.
Six stages, in this order.
We start by understanding the decision the system has to enable or the operation it has to speed up. We leave this stage with a written scope, the known constraints, and what is explicitly out of scope.
Data model, component boundaries, technology choices and infrastructure sizing. Rejected alternatives are recorded with the reason, because the question will come back in a year.
Before building everything, we tackle the most uncertain element: a model's accuracy on your real data, a video stream holding up under load, an integration with a third-party system. If it does not work, now is the time to find out.
We deliver working sets of functionality you can try, not percentages of progress. Every increment is deployed to a test environment you can reach.
Hardened servers, certificates, encrypted backups with a tested restore, monitoring and alerts. Deployment is scripted, therefore repeatable.
Code, operations documentation and credentials are handed over, with a transfer session for your team. We remain available for maintenance if you want it, not because you have no alternative.
When the need arrives as a tender pack rather than a brief we write together. The order below is the usual one; the buyer's rules take precedence.
We read the consultation file, the criteria, the deadlines and the required references. If the need or the means asked for are not something we know how to deliver, we do not bid.
When the pack is ambiguous, we put written questions within the stated deadline. The buyer's answers bind every candidate; we do not treat them as a private channel.
If a lot goes beyond what we cover alone — a domain we do not practise, a local presence, production capacity — we join a consortium or lead one. That is a decision, not a reflex. The lead firm and the split of lots are written down before submission.
We submit a written offer: method, team, schedule, price. We do not match a price we could not hold, and we do not write a method we would not carry out.
If the procedure includes a hearing, we present the offer as written. It is not the moment to invent an option that is not in the file.
Awarded or not, the outcome is the procedure's. If we are awarded the contract, we open the same project scoping as for a privately negotiated contract, rather than starting to build from the offer alone.
A scope to challenge, an architecture to validate, a system to take over — start by writing to us.