The organisation needed one AI creative workspace for creators, reviewers, and workspace administrators. The existing journey divided creative briefs, prompt history, image generation, asset versions, collections, team review, usage limits, and approved exports between separate tools and informal follow-up, so the project had to solve the operating model as well as the interface.
A complete AI creative workspace shaped around creators, reviewers, and workspace administrators
The organisation needed one AI creative workspace for creators, reviewers, and workspace administrators. The existing journey divided creative briefs, prompt history, image generation, asset versions, collections, team review, usage limits, and approved exports between separate tools and informal follow-up, so the project had to solve the operating model as well as the interface.
Discovery followed the full journey from the first customer or staff action to its operational and commercial close. That exposed the handoffs between creative projects, asset review, and ai generation, including the exceptions that a feature list would have missed.
- Map the people, decisions, and records across the whole AI creative workspace.
- Separate firm creative technology rules from habits created by the old tools.
- Agree what the first release must complete from end to end.
The product joined creative projects, asset review, and ai generation without blurring roles
Treat every generation as a versioned project asset, separate experimentation from approved work, and make cost and model choices visible.
In the AI creative workspace, each audience receives a focused home while the underlying state remains shared. The interface shows what needs attention, who owns the next action, and why work is blocked, while secondary detail stays available for review without crowding everyday tasks.
- Design the primary journey for every audience named in the brief.
- Give exceptions, corrections, and cancellations an explicit route.
- Use realistic records to test the complete flow before widening the surface area.
React, Python, Object storage formed one supportable AI creative workspace
The architecture follows business ownership rather than the navigation menu. React, Python, Object storage share versioned contracts, server-side permissions, observable integrations, and one authoritative record for the decisions behind asset review.
In the AI creative workspace, slower or external work runs through durable jobs with idempotent handling and visible failure states. Logs retain the organisation, actor, and product record needed to investigate a problem without guessing from a generic error message.
- React supports the primary product surface.
- Python owns a bounded part of the product journey.
- Object storage keeps operational data and handoffs reviewable.
The release had to prove the whole AI creative workspace, not one polished screen
Creative teams can explore quickly without losing the brief, review history, or final approved assets. The release plan therefore followed representative work across every audience and the complete path through creative projects, asset review, and ai generation.
Acceptance for the AI creative workspace included ordinary work, missing inputs, repeated requests, failed integrations, permission boundaries, and the route back to a trustworthy state. The team checked whether staff could understand and recover from a problem without direct database help.
- Run the busiest realistic journey from start to operational close.
- Rehearse a failed integration and a repeated customer action.
- Confirm that every role sees only the work and data it owns.
- Trace a management result back to the records that produced it.
What the work established
Creative teams can explore quickly without losing the brief, review history, or final approved assets.
- 01
The AI creative workspace now gives creators, reviewers, and workspace administrators one shared operating record.
- 02
Creative projects, asset review, and ai generation follow the same product rules.
- 03
The system makes ownership and exceptions visible instead of hiding them in manual follow-up.
- 04
The product can grow from clear workflow, data, and service boundaries rather than isolated features.