Story Hub.
From a screenplay to shared understanding. Designing an AI partner for the people who make the story.
The production workspace · Story Hub
The problem
Every team read the same screenplay, then rebuilt its meaning in separate documents and conversations.
The solution
One story-aware workspace. Explore the script, follow the evidence, and develop creative directions with the team.
The impact
Weeks → hoursScreenplay coverage in a 12-title pilot. The vision secured a funded, multi-year program.
- My role
- I led design for the AI pod from vision through launch: interaction design, grounding, citations, and product quality. I partnered with PM, engineering, research, and eight partner organizations.
- The central decision
- Make interpretations inspectable. Keep the screenplay, supporting evidence, and creative controls close to the answer.
- Project scope
- A 12-title pilot, a funded Media 2030 program, and explorations across the wider creative workflow.
The decisions behind the experience
Watch the interaction. Follow the decision.
The open screenplay sets the context. Citations connect an answer to its source, keeping the conversation and evidence side by side.
Start from the story, compare possible directions, then refine the details. Creative judgment stays with the team at every step.
One screenplay. Eight organizations. A different job for each.
The screenplay connects development, production, marketing, and release. But the understanding around it was scattered. Every team opened the same document, then rebuilt its meaning in spreadsheets, decks, notes, and conversations.
For an executive, the question might be whether a story is ready to greenlight. For production, it is what a rewrite changes. Marketing needs the story’s themes and audience. A useful AI experience had to recognize these jobs while preserving one shared source of truth.
The ecosystem and the people in it
01 / 04
The obvious answer was a chat box. The useful answer was a collaborator.
I explored multi-script conversations, persistent sidebars, global chat, and more autonomous agents. Paper sketches and mid-fidelity frames let us compare density, panel structure, and what happened when AI competed with the script for attention.
The work shifted when a content partner pushed back on AI interpreting the story for them. People wanted help examining possibilities, while keeping ownership of what their story meant. I reframed the product around collaboration, evidence, and human authorship.
From exploration to a working model
01 / 06
My role beyond the screen
I established grounding rules, citation behavior, and multi-answer patterns, and helped create Netflix Studio’s first internal AI interaction guidelines. Weekly critiques across four pods kept the emerging experience coherent.
“Working with Michael has been one of the most collaborative partnerships I’ve experienced. He shares his thinking openly, welcomes pushback, and treats every project as a team effort.”
Before AI creates anything, it needs to know the story.
The first layer made a screenplay easier to interrogate. Script chat inherited the open document. Executive summaries, breakdowns, and connected story data carried citations back to the page. People could inspect the evidence without losing the conversation.
That relationship shaped the interaction architecture: user intent routed to the right specialist, with scripts, notes, documents, and history held in a shared context layer. The experience needed to feel like one continuous collaboration.
A grounded understanding layer
01 / 05
Trust is an interaction, not a disclaimer
Citations sit beside the answer. Context carries across tasks. People can inspect the material, question an interpretation, and decide what happens next.
Give the team more possibilities and the controls to shape them.
Once teams had a shared understanding, we could build a creative layer on top of it. Character studies, moodboards, storyboards, and budget scenarios begin with the screenplay rather than an empty prompt.
I designed for branching and refinement. A generated image is a starting point: select a region, make a specific change, compare directions, and bring the work into a shared canvas. That keeps creative judgment with the team.
From the page to a creative direction
01 / 07
A workspace that remembers the work, wherever it happens.
Home, productions, assets, and team spaces give the intelligence a product to live in. Productions carry their own identity and activity. Returning to the workspace means returning to the story, rather than rebuilding context.
I also explored continuity across mobile and existing Studio tools. Story intelligence could follow someone from the production workspace into scheduling and creative review. The longer-horizon concepts below are explorations of that direction, distinct from the launched capabilities.
The product around the intelligence
01 / 12
Future concepts · exploring the next chapter
01 / 06A faster understanding. A funded direction.
In the 12-title pilot, screenplay coverage moved from weeks to hours. The platform connected eight partner organizations and more than 350,000 screenplays. The working prototype also created organizational momentum: a funded, staffed, multi-year Media 2030 program.
The timing result comes from the 12-title pilot. The screenplay count describes the platform’s broader collection.
I start with what the experience should remember, what people need to inspect, and where they should take control. That interaction model can carry new capabilities long after a particular screen changes.
