2. Camille
Camille
The dashboards update every three seconds. I don’t need them to. Nothing meaningful changes at that frequency, but the illusion of movement keeps Eon executives calm, and calm executives are less likely to make decisions that require me to fix them later.
So I let the refresh cycle run.
Top screen: live sentiment analysis. Second: engagement distribution by region.
Third: flagged anomalies, currently filtered to “high-risk viral potential.” Draft responses sit queued beneath them, pre-approved language ready to deploy.
Statements, denials, reframes. I don’t write them from scratch anymore. I refine.
Everything is color-coded. Everything is contained.
This is what control looks like.
I oversee narrative containment for Eon. When something supernatural becomes visible, the Council deals with the threat. My department decides what the public believes it saw.
I take a sip of my coffee and skim the overnight reports again. I’ve already read them twice, but repetition catches what urgency misses.
Harbor’s Edge is still sitting at the top of every feed.
Not the incident itself. That’s been buried under three layers of acceptable explanation, two coordinated press releases, and one carefully timed influencer campaign that I personally directed. But the echo of it. The shape of it.
Residual engagement is higher than projected.
That’s not surprising.
People don’t forget something that almost felt real. They just need help deciding what to call it.
A flagged clip blinks in the corner of the screen.
Harbor’s Edge Incident — Unverified Footage.
I open it.
Grainy. Vertical. Shot from too close to be safe and too far to be useful. There’s movement at the center, too many points of articulation, not enough coherence. The audio is worse. Layered, distorted, like multiple voices trying to occupy the same space without agreeing on the order.
The Chorus.
I watch exactly three seconds, then tag it: anomaly, escalation risk, containment failure.
Time to suppression: four minutes, twelve seconds.
Unacceptable.
Not catastrophic. But not acceptable.
I queue a takedown request before the timer finishes calculating, flag the originating account, and route the clip through suppression channels. By the time I’m done, it will be gone, replaced, or rewritten into something safer.
The problem is not the incident.
The problem is that it was seen.
And it keeps being seen.
New accounts. Reuploads. Compilations. Slowed-down clips with captions that ask the right questions in the wrong tone.
Midnight Halo is trending for the wrong reasons.
The initial response should have contained it. It didn’t. By the time the fallout reached my department, someone had already decided a denial and a few removed videos would be enough.
The interview didn’t help. Jules never does.
I pull up the engagement threads again and watch the pattern form in real time. Interest becomes narrative if it’s left alone, and whoever handled the first response treated Harbor’s Edge like a simple incident.
It isn’t.
It’s a story.
And right now, it’s being written without us. That’s the problem.
Someone knocks once. Not tentative or formal, just enough to register.
“Come in,” I say, without looking up.
Kevin steps inside like he belongs there, which, technically, he does.
Kevin is technically my assistant, though he’s been with Eon Entertainment longer than I have. He understands the rhythms of this place better than most and remembers what this department looked like before I took it apart and rebuilt it.
Physically, there is nothing remarkable about him.
Average height. Soft around the middle in a way that suggests more desk lunches than gym memberships.
Brown hair beginning to thin at the temples, and posture slightly rounded from years spent leaning over other people’s problems. He looks like someone’s dependable suburban father who wandered into the entertainment industry by mistake.
He sets a cup on my desk. I don’t need to check it. I already know it’s exactly right.
That’s the remarkable thing about Kevin: he pays attention to everything, and he remembers everything. Attention to detail is something I respect. It’s also why he’s still here.
“Long morning?” he asks.
“It’s no longer morning,” I say.
He glances at the wall of screens. “Well, they say time is a construct.”
“Not for scheduling,” I reply, scrolling through another report.
He hums like that’s a philosophical stance instead of a practical one. “You skipped breakfast.”
“I didn’t,” I say. “I chose not to eat.”
“That’s not better.”
I do not dignify that with a response. He leans against the edge of the desk, not interfering, just… present. Kevin does that. He occupies space without demanding it.
“You’ve been in here since—“ he checks his watch, ”—too long.”
“That’s not a metric.”
“It should be.”
“It isn’t.”
He studies me for a second, like he’s deciding whether to push. He used to. He stopped pushing after the Valencia incident. Everyone did.
My fingers find the clasp of my watch and straighten it, though it is already aligned.
“Anything I can do?” he asks instead.
“Yes,” I say. “Make sure no one else brings me coffee. This one is correct. The next one will not be.”
“Can do.” He straightens, already halfway out the door. He pauses just long enough to catch my attention. “Camille?”
I glance up. There’s something else there for a second. Not hesitation. Consideration. Like he’s deciding whether this is the moment he says something he shouldn’t.
“Take a break at some point.”
“I will,” I say.
I won’t. He knows that. He leaves anyway. The door clicks shut, and the room settles back into its natural state of quiet efficiency.
A notification slides across my screen. Priority flag. Internal. I open it without hesitation.
Midnight Halo performance data populates across the main display.
Every scheduled appearance is assigned a performance window: the block of time between the first live cue and the approved close.
Production uses the term to contain every measurable variable within a defined interval: audience response, broadcast exposure, security access, and projected revenue.
Midnight Halo’s highest volatility occurs inside those windows.
Tour projections already exceed last quarter by twelve percent. Audience retention is high. Engagement spikes correlate cleanly with choreography peaks, vocal drops, and emotional cues embedded in set design.
It’s a good system.
Heat maps bloom across the screen, color-coded emotional response over time. Red for intensity. Blue for drop-off. Yellow for sustained engagement.
The goal is not constant intensity. That burns out too quickly.
The goal is controlled peaks.
Build. Release. Build again.
I scroll through the performance logs, tracking where the audience reacts, where they don’t, where attention spikes beyond projected ranges. High emotion correlates with high retention. High retention correlates with revenue.
This is not complicated.
It’s just rarely managed correctly.
I open the member profiles.
Midnight Halo — Internal Use Only.
Kaia: leadership stability. High control. Reliable under pressure. Low volatility. Strong anchor point.
Remy: tactical reliability. Pattern recognition. Low emotional variance. Useful for internal stabilization.
Mina: emotional sensitivity. High empathy. Reactive to environmental shifts. Potential liability. Also potential amplifier.
I move on to Jules. Her file is thicker. More data points. More fluctuations. More… everything. Engagement metrics spike immediately. Hers are the highest across all platforms. Audience response curves peak faster, hold longer, drop harder.
Jules: elevated volatility index. High-value asset. Severe performance-window instability. High containment priority.
I pause on that.
Not because it’s surprising, but because it’s accurate.
Jules both drives and destabilizes engagement.
Both are useful. Both require management.
She’s a variable. A high-impact one. Which makes her priority.
My screen flashes with an incoming call from Alaric Voss. Executive Oversight.
I recognise the line immediately. He doesn’t call unless something has already escalated.
I accept.
The voice on the other end is smooth, measured, deliberately neutral. “Camille.”
“Yes.”
“We’ve reviewed your initial assessments.”
“I’ve sent three versions,” I say. “Which one are you referencing?”
“The one where you flagged Halo as a potential containment risk.”
“I did not say potential,” I reply. “They are a containment risk.”
A pause.
“We need stability.”
“You have stability,” I reply. “These are perception issues.”
“Which is why you’re being reassigned.”
“Define scope,” I say.
“You’ll be embedded with Midnight Halo for the duration of their world tour.”
Embedded. Not consulting. Not advising.
Managing.
How inconvenient. How inefficient. I’m better at directing correction, not performing it in real time. Babysitting talent is not a productive use of my time.
“Primary objectives?” I ask.
“Maintain profitability. Prevent further leaks. Control narrative drift.” But then there’s a slight shift in Voss’s tone as he adds, “Secondary objective: Jules Navarro.”
I don’t interrupt.
“She’s high exposure,” Voss continues. “Her last interview destabilized multiple channels. We can’t afford a repeat during the tour. She requires active management. Control her messaging. Limit unscripted interaction. If she deviates, you correct it immediately.”
“Understood,” I say.
“Upper oversight is watching,” he says. “They expect continued performance without disruption. We cannot afford a public misstep. Not now.”
“I’ll handle it,” I say.
“We expect you will.”
The line disconnects.
I sit back for a moment, just processing. The system isn’t broken. It’s under strain. There’s a difference. Strain can be managed. Mismanagement can be corrected. Narrative can be redirected.
I don’t fix people.
I fix perception.
Jules appears on screen mid-routine, movement sharp, controlled, precise. Predictive metrics overlay the footage in real time, modeling audience response against timing, focus distribution, and choreographic emphasis.
The curve spikes early.
I adjust the model and run it again.
Same result.
Interesting.
I watch her again.
She pulls attention before the expected cue and holds it past the drop. The model accounts for positioning, choreography, vocal emphasis, and screen time. The curve should level there.
It doesn’t.
Charisma explains attention.
It does not explain a predictive model repeatedly failing around one person.
That makes Jules an anomaly.
I’ll start with the most volatile element.
Jules is where the system breaks, so she’s where I’ll begin.