a show is one graph, and everyone was looking at a different corner of it
a production of that size runs on spreadsheets, decks, pdfs and group chats, and each discipline keeps its own copy of the truth. the lighting plot says one thing, the labour call another, the credential list a third. the work of a lead project manager is mostly reconciling them by hand, at two in the morning.
we wanted the show to be one graph, with every document a view of it.
one graph, a portal per persona
the production graph holds the venue, the stages, the rig, the schedule, the people and the paper. around it sit scoped portals: the promoter sees the advance and the capacities, the artist team sees their own day and nothing else, the vendor sees the rfp and their bid, the crew sees the call. publish is a model, not a button: draft, impact, publish, with every export generated from the graph so a change lands everywhere at once.
- the schedule brain: the org axis, rig-line editing, the plan as the backlog
- credentials with a date on every comp, routes and the venue map
- the artist advance, per act, generated from the graph
- bids and rfps per discipline with extraction lanes for the replies
- the labour instrument and the call
- the spec studio and the warehouse
- publish everywhere: draft, impact, publish, exports with zero re-keying
- a security gate on every change: gitleaks and semgrep in the pipeline
built in the gaps of the show it was built for
the first commit landed at the end of july. by august the ui had its second version and every noun in the catalogue had become a door. september was the publish model, the security gate and the overhaul sprints for the show: the credential model, the lighting re-cut to the final rig, the venue map, the close-out across fifteen personas, the phone layout.
it is two:am's own product. the show was its proving ground, and it will carry the next one.