Be the director: how stage craft informs product craft ft. Kim Beverett (OpenAI) | Config 2026

Hello everyone. Welcome back to the L2 stage.

Hi, welcome back. It's great to have you. My name is Majola.

you. My name is Majola.

I'm Isaac. And it and I hope you had a great ch or I hope you had a chance to grab a snack, maybe a stretch because we're going to get right back into it on the level stage on the level two stage.

Sorry.

That's right. And our next talk, I'm so excited because we are about to learn how stage craft informs product craft.

Please give a warm welcome to Kim Bever.

Let's do it.

Today I'm going to give you a crash course in theater. And hopefully you're going to leave here feeling like there's very little difference between this and

this. They both frame the audience's

this. They both frame the audience's attention. Both say focus here. forget

attention. Both say focus here. forget

that world and enter this world. And

within these two rounded rects are near limitless potential. My name is Kim

limitless potential. My name is Kim Bever. I am a product manager at OpenAI.

Bever. I am a product manager at OpenAI.

I've been a founder. I was at Apple for 10 years and I am authorized to give you this crash course on theater because I have a degree from the illustrious Theatrical Institution, Arizona State

University.

I've been an actor, a director, a stage manager, a costume designer, and obviously I wasn't exceptionally good at any of those things. Otherwise, I

wouldn't be here giving you a talk today. But I was and still am obsessed

today. But I was and still am obsessed with one thing, telling stories. People

asked me how I moved from theatrical design to product design, and I tell them I never stopped designing for the theater.

I just design for a different audience now.

So, let's kick off our crash course.

This is a chair.

Imagine you walk into a theater and you see this chair. What do you think about the show you're about to see? Probably

that it's pretty standard standard, straightforward, not a lot of surprises.

Now, you open an app and the first screen looks like this. Remember, the

user is your audience. What does that audience think? Probably this is a

audience think? Probably this is a pretty standard, straightforward app.

Not a lot of surprises. Now you walk into this theater.

What do you think? That's weird. What is

that chair doing? Why is that chair there? That chair's over the side. What

there? That chair's over the side. What

is that? Do you see that chair? Now I,

the audience member, I'm obsessed with that chair. And if you wanted me to be

that chair. And if you wanted me to be obsessed with that chair, congratulations. Job well done. If you

congratulations. Job well done. If you

didn't, you now have a problem.

Same thing here.

What is that button doing? I'm obsessed

with it and it hurts me. This is the uncanny valley of choices.

Now you walk into this theater.

Now I'm excited. That is not normal chair behavior, but I know you did it on purpose because you wanted me to be interested and engaged from the moment I

sat down.

Same thing here.

Weird choice.

Clear choice. It's a choice that makes me pay attention and listen. It's a

choice that a director would make because committing to choices is a director's primary job.

There's a story from a director I love, her name is Anne Bogart, and she tells this story about a young director in a really busy rehearsal and an actor comes

up and goes, "Where do I put this chair?" And in her haste, the young

chair?" And in her haste, the young director says, "It doesn't matter."

But in theater, as in software, we know that's wrong because to an audience, to the user, everything matters. When an

audience member sits down in that seat, they assume that everything they're about to see was an intentional choice.

It's tempting to look at this and see it as like a lack of choice, as some sort of representation of neutral, but it's not neutral. Within 5 seconds of

not neutral. Within 5 seconds of entering this theater, I know the director chose to leave the curtain open. So, I'm going to be thinking about

open. So, I'm going to be thinking about what's about to happen. They used a plain kitchen chair, not a fancy chair, so I'm going to expect minimalism. They

chose to light the center of the stage only, so I'm going to be a little uneasy wondering what's happening outside the field of view. There is no neutral. Your

audience is sophisticated. They feel the results of your choices, even if they couldn't explain why. The audience came to the theater not to see any show. They

specifically came to see that director's choices, your choices.

Everything is a choice.

That's it. That's the first lesson. So,

be intentional with your choices because the audience assumes you were intentional with your choices. Okay,

that was the first lesson. You ready for the advanced lesson?

I'm about to show you several photos from one of the most groundbreaking shows of the 20th century. And we're

going to try to guess what show it is. I

promise it's a show you've heard of. I

also promise it's a little silly and you're allowed to laugh. Okay.

We got a big old red feather. We got

some tie-dye shirts in swings, trapezes to hammocks, guy on stilts. Uh, more tie-dye people

running around that. Um, we got a trappies and a big white box.

Okay, I promised it was a show you've heard of. Anyone know what show this is?

heard of. Anyone know what show this is?

This is a Midsummer Night's Dream by the great Shakespeare himself, directed by Peter Brookke at the Royal Shakespeare Company in 1970. And as we learned in

our first theater lesson, good theater is made up of strong choices. And this

show is a lot of strong choices. But

first, a quick plot recap. Heria loves

Lander. Helena loves Demetrius. Heria is

told to marry Demetrius instead. She

says, "Hell no." Runs to the forest with Lzander. Demetrius chases Hermier.

Lzander. Demetrius chases Hermier.

Helena chases Demetrius. In the forest, there are fairies. They do fairy stuff.

The fairy king gives them a love potion so they love the wrong person. One guy

gets turned into a donkey. They all wake up and laugh it off. Typical

Shakespeare.

But the staging and costumes here are anything but typical. At the time, A Midsummer Night's Dream was seen as mostly unserious. When people did stage

mostly unserious. When people did stage it, they leaned into the magic and the fairies and the forest and the silliness, and it was a play for kids.

If you want to see what that looked like, you don't have to go far for an example. This is a Midsummer

example. This is a Midsummer Nightstream. Both of these, the exact

Nightstream. Both of these, the exact same scene when the fairy king gives the love potion by the same theater company just 8 years previously.

Again, same production, eight years different. And you can see what a

different. And you can see what a difference eight years makes for a donkey costume.

Also, fun fact, uh, in the black and white, that's Judy Dench playing Tatana.

So, Peter Brookke, our avantgard director, he does not do theater like you see in this black and white picture.

So, why would he consent to do a show like Midsummer Night Stream? Because he

knew that the script is not the show.

The show is the choices layered on top of the script. There's a reason Shakespeare is performed a thousand times. We go back again and again

times. We go back again and again because we want to see the interpretation, the perspective, the choices that a new director can bring.

It's the same reason there's a cajillion to-do list apps. They've all got the same basic script, but we keep coming back because we're hoping for a new layer of choices that's actually going

to help us return our library books.

Brooke wanted to put a very specific layer on top of his Midsummer Night stream. He wanted to go beyond

stream. He wanted to go beyond illustration to evocation. He believed

that evocation was more powerful than illustration, which frankly is a really bold choice. You're doing a show about

bold choice. You're doing a show about magic and you don't want to illustrate it. No fairy wands, no sparkles, no

it. No fairy wands, no sparkles, no glitter. No, just a simple white box

glitter. No, just a simple white box that lets you project your imagination onto it. Because the minute you put a

onto it. Because the minute you put a real forest on stage, that's the minute the audience's imagination turns off.

There's no need to work. There's a

forest right there. In this picture, these wire balls, they represent bushes and foliage in the forest. Brooke gave

us nothing familiar to latch onto. He

forces the audience at every turn to use their imagination because Brooke believed evocation was greater than illustration. And that was the spine of

illustration. And that was the spine of his show. And the spine of a show is a

his show. And the spine of a show is a crucial theatrical concept.

A spine is a single overarching sentence to which every choice must adhere.

Brook's spine was evocation over illustration. But if you were directing it, you could do something else like the falsehood of choice or the ritualistic nature of humans.

It's less important what the spine is and more so that it exists and that it resonates with the director and the cast and the crew. The spine is what allows

diverse choices to tell the same story.

What allows teams of radically different disciplines in the theater to function.

Costumers, sound designers, actors, the director's primary job is to unite those choices under the spine of the show. We

can see this in the set of Dream. this

iconic white box. This is not credited to Brooke. It's credited to set and

to Brooke. It's credited to set and costume designer Sally Jacobs. This, by

the way, is the actual model that she built with her hands to pitch this idea because there was no Figma.

Because Brooke set such a strong spine, Jacobs could make this crazy choice.

This groundbreaking work didn't spring from the life of a single-minded genius man. Brooke admits that when he signed

man. Brooke admits that when he signed on for the production, he had a burning hunch that somewhere an unexpected form was waiting to be discovered,

which means he didn't have a plan. He

had a script, a spine, and 10 weeks to figure it out, which by the way, absolutely unheard of, 10 weeks. At the

time, the Royal Shakespeare Company was giving productions three weeks to rehearse. But those three weeks or 10

rehearse. But those three weeks or 10 weeks gave them the chance to experiment.

They got to bring an audience in to watch rehearsals and that audience informed their choices. You can see here Sally Jacob's first drawings of the workmen including the donkey. And what a

journey they went on from the original drawing to the final production. Brooke

knew that theater can only exist with the participation of an audience. A

beautiful painting can exist without a viewer, but theater cannot.

If good theater hinges on your choices, then you have to validate those choices as early as possible. The audience is not a passive viewer. They are a collaborator, a mirror, a witness.

Brook's dream could not exist today because the audience from 1970 doesn't exist today. Because we've all seen a

exist today. Because we've all seen a white box in abstract costumes a million times. And it doesn't trigger our

times. And it doesn't trigger our imaginations in the same way. Entirely

new choices would be required to support that same spine because we, the audience, the final members of the show, are different.

This skumorphic choice made a lot of sense in 2009. Users were totally unfamiliar with a touchscreen and apps and all these paradigms, and they needed

the handholding. Today for us, not so

the handholding. Today for us, not so much.

This wacky 70s play with white walls and barbwire bushes and trappies magic changed the theatrical world. Without

this play, we don't have Lion King or Hamilton or any of the theater we're accustomed to.

And now to bring it back to software. I

am blessed to work with a team that feels the same way as I do about software, even though they probably don't have any idea that I think of our features as little software plays. Um,

for example, we recently released a new feature into codecs. It's called

computer use. It gives your computer a cursor and a keyboard so it can take action in your apps in the background.

But we had to figure out how to represent that action taking to the user. We needed to show a cursor. So, we

user. We needed to show a cursor. So, we

could have just done this. Your computer

already has a cursor. Just move that sucker from point A to point B and job done. That was the product requirement.

done. That was the product requirement.

That's what the script necessitated.

But instead, we did this. We gave it a little glow and some personality and a little dance along the way. And this was hard. Our designer customuilt this

hard. Our designer customuilt this simulator to mock up all the potential paths the cursor could take. The

engineer working on this spent days getting it right and lost sleep in some corner cases where the cursor didn't land just right.

But why? Why spend all this extra effort?

Because one choice says it doesn't matter and one choice says everything matters.

We've also got a team that loves to rehearse. We just released another

rehearse. We just released another feature called appshots that lets you easily pull in context from your apps into codecs. And I can show you what it

into codecs. And I can show you what it looks like today. So you just press command command boop and the context of your app, your screenshot and text is pulled into

codeex so you can start working on it.

But it took a year to get to this. I'm

going to show you the journey. It's a

little chaotic but stick with me. First,

at the top, we tried adding the users's apps to the chat bar. Didn't work. Then,

we tried magically inferring what app you wanted and letting you include it.

Didn't work. Then, we tried this lasso to connect them. Didn't work. But we got to the right solution with rehearsals, with working with our audience to find

what actually made the task easy.

Both of these features tell you a lot about the team that made them. The team

that made these feels that computers are fun. That we are living in a time where

fun. That we are living in a time where software is producing near magic, but it rarely feels magical.

I do not bring up these examples to say that my team is building the Midsummer Night's Dream of Far from it. I just

bring these up to say that even in the smallest of examples, you can make choices that help tell your story. So,

at this point, I hope that I've sold you on the value of choices. Easy. Make some

choices, right? Wrong.

When I first previewed an early version of this talk to Anna, my exceedingly kind and patient handler for this config talk, she said,

"I think this is probably the most horrible and important piece of feedback I have ever received. Me, the person standing on stage making a passionate

plea for showing your choices in your work." And I wasn't doing it because I

work." And I wasn't doing it because I was scared not of showing a bunch of fancy designers avantguard '7s theater although that is terrifying. I was

scared of sharing the real spine of this talk. I was scared because making these

talk. I was scared because making these choices is scary on their own. Each

choice closes the door, obliterates optionality.

The violent act of choice is scary, but it's doubly scary when you're asked to make those choices based on your own perspective and lived experience because

that's what theater is. It's a brave plea for shared experience. You remove

the costumes and remove the lights and it's one person telling a story to another person. It's saying, "Here,

another person. It's saying, "Here, this story is too big to hold inside my mind and my body. Will you help me hold it?"

it?" It's taking the time to tell your story with purpose and care such that every good show holds that warm echo of human

intentionality.

It's hard to think of something braver than that.

But for years, we have layered rules on top of this brave act. You've probably

heard of these. Aristotle was dictating how we tell stories in 335 BCE. He came

up with this stackranked, yes, stackranked list of six crucial elements of theater. What happens, who does it,

of theater. What happens, who does it, how it sounds, and these rules still persist today. That last one, spectacle,

persist today. That last one, spectacle, that's like the wow of a show. That's

just surprise and delight. Steve Jobs

knew his classics, but I don't remember the stories that adhere strictly to those rules. I remember the stories that

those rules. I remember the stories that break them.

What is maddening about our current moment is that we are building software like these rules are set in stone.

We're using the most extraordinary technology that most of us have ever seen to create terribly ordinary work.

Work that is the average of all of humanity's collective choices with the fingerprints removed and the edges filed off.

This prompt is never going to produce something incredible.

We cannot expect the averaging machine to produce anything other than averages.

Imagine with me. A director walks into the first rehearsal.

Make me a musical about Alexander Hamilton. Funny with some rap.

Hamilton. Funny with some rap.

No.

You want to see what that prompt made?

Perfect.

No notes.

Now, there is nothing wrong with using these tools. Half of the images in this

these tools. Half of the images in this talk wouldn't exist without them.

I am such a big fan of these tools that I tried really, really, really, really hard to have AI write this talk for me.

And you know what it got me? It got me Anna saying that she felt like I wasn't bringing myself to my talk. Because try

as I might, I could not get the tool to understand my unique human experience. I

guess there just wasn't enough costume designer turned product manager in the data set.

The spine had to come from me.

Brook's Dream premiered in 1970 and it caused a ripple in the theater that we still feel today. But in 1969,

no one saw it coming. And right now we are in a 1969 moment.

We are on the eve of a software and design revolution that will ripple out for generations. A radical new form of

for generations. A radical new form of software is coming. And we are not going to get there with this.

We are going to get there when we make choices that show who we are. When our

choices show that warm echo of human intentionality that is going to dictate the new forms we discover.

Brooke himself was aware of this need for change. The great tragedy of his

for change. The great tragedy of his Midsummer Night's Dream is that it was never properly filmed. And when asked why, he said this.

No form nor interpretation is forever. A

form has to become fixed for a short time. And it has to go. As the world

time. And it has to go. As the world changes, there will and must be new and totally unpredictable dreams.

It's time for new dreams, new forms, and I cannot wait to see what dreams you create. Thank you.