One roster to staff from, and no artist sees another’s rate

• A list, not a network. Plenty of artists were known to the studio, but there was no current, searchable, checked record of them in one place.
• A spreadsheet for every kind of vendor. Individual freelance artists were tracked in one set of sheets and vendor companies in another, each maintained by hand, so there was no single place to look and no way to tell which copy was current.
• Nobody owned the data. Details went stale because keeping them current was somebody else's job, and a spreadsheet has no way to ask an artist for a new phone number.
• Casting by hearsay. Finding the right person for a sequence meant asking three colleagues and hoping, on work that needed filling immediately.

The objective. One network the studio can search with confidence, kept current by the artists themselves rather than by the people hiring them.

The original brief specified thirteen modules across twenty-one screens. Rather than build all of it at once, we recommended phased delivery, starting with the one module every other part of the brief depended on.

• Core first: the talent network and review workflow. A roster nobody has filled in is worth nothing, and every other module in the brief assumed one that was populated and current, so that was built and put into use first.
• A documented roadmap for the other twelve modules. Each was specified and scheduled rather than quietly dropped, so the studio knows what comes next and in what order.
• Scope settled before the estimate. Fordawn recommended building the core first before giving any estimate, rather than quoting for the whole list and discovering the dependency later.
• Interactive prototypes for desktop and mobile. Every screen was designed as a clickable prototype before development, with the phone treated as its own design rather than a scaled-down desktop layout.
• Annotated for sign-off. Each screen came with notes naming every control and what it does. The studio walked through the prototype, chose the visual direction and sent back its own ideas before any code was written.

• Mobile-first onboarding. A busy freelancer completes the profile once, usually on a phone, in a single sitting. The form is built so that work is not lost partway through, because an artist who loses their progress rarely comes back to start again.
• Locked records during review. Once submitted, a profile cannot be edited while it is under review, so the reviewer assesses a fixed copy that cannot change underneath them.
• Concurrency-safe decisions. When two reviewers act on the same application at the same moment, only the first action is applied. The second reviewer is told it has already been handled. An artist never receives two conflicting emails.
• A permanent audit trail. Every decision is recorded with who made it and when, and the record survives account deletion: remove an artist entirely and the log still shows who removed them, when and why.
• Network-wide search and filtering. Search runs across every accepted artist, not just the fifty on screen. Filters cover discipline, experience level, availability, and visual effects or AI. Results sort by name, experience, rate or date.

No artist can see another artist’s rate, contact details or documents. That rule is enforced in the storage itself rather than by hiding fields on the screen. Five protections keep it true:

• Four layers, each of which holds on its own. The site turns away anyone who is not signed in, each screen checks again for itself, every action checks a third time before it runs, and the storage underneath refuses regardless. This is defence in depth: any one of the four failing changes nothing.
• No code from other companies. Many sites quietly load scripts from advertising and analytics vendors, and each one is a way in. The artists' portal loads none of them and talks only to itself. When a third-party service was proposed for one feature, it was declined for that reason and the equivalent was built in-house.
• Password guessing does not work. Repeated attempts are shut down quickly, which stops credential stuffing, the practice of working through a list of stolen passwords. The limit is tuned so a whole office sharing one internet connection is never locked out by accident.
• Keys are checked before every change. Automatic secret scanning runs on every change before it ships, so a password or access key committed by accident is flagged before it reaches the live system.
• The password reset never reveals an account. The forgotten-password page gives the same answer whether or not an account exists, so it cannot be used to find out who has one.