[Gardner Analytics Office — August 2014, Morning]
Marcus's news aggregation script — the same one he'd built during his second week, now expanded to monitor a dozen sources in real time — pinged at 8:47 AM with an alert flagged PRIORITY: HIGH.
"Hooli Names Vincent Mora as VP of Machine Intelligence, Signals Major AI Push"
The article loaded on Marcus's monitor. Sarah saw it first from across the room — her desk faced his, a configuration she'd arranged so they could exchange code reviews by turning their screens toward each other.
"Ethan," Sarah said. Not a shout — the particular inflection she used when delivering information that required immediate processing. The same tone she'd used when Marcus Webb's blog post appeared. The same tone from the morning the ChronoCloud invoice had been flagged.
Ethan crossed the office. The expanded space meant twelve steps instead of four, a luxury he was still adjusting to. The article filled Marcus's twenty-four-inch monitor.
Vincent Mora. The name surfaced in Ethan's memory — not from the show, where he couldn't remember a character by this name, but from the tracker his mind automatically maintained on Hooli's movements. Vincent had been mentioned in the internal intelligence Sarah had gathered after the HooliBot disaster: Gavin's new VP, hired from Google, tasked with rebuilding Hooli's AI credibility from the ashes of HooliBot's public failure.
But the press release painted a different picture from the quiet hire Ethan had expected. This was a public statement. A declaration.
"Hooli is committed to leading the next generation of artificial intelligence," Mora said in the announcement. "The future of computing is intelligent systems that understand, reason, and create. We will build that future at a scale no other organization can match."
The language was precise. Not Gavin's empty buzzwords — Mora's quote used "understand, reason, and create" with the specificity of someone who knew what those words meant in a machine learning context. "Create" was the dangerous word. It implied generative capability. It implied awareness of what Gardner Analytics was building.
"He's from Google?" Ethan asked, pulling up Mora's LinkedIn on his own laptop.
"Google Brain," Marcus confirmed. "Seven years. His publication list is... real. Recurrent networks, sequence-to-sequence models, attention mechanisms in translation."
Attention mechanisms. The two words sat in the air with the weight of a loaded weapon.
"He knows what we're doing," Sarah said.
"He knows what attention is. He doesn't know what we've built with it."
"Yet."
Ethan studied the LinkedIn profile. Stanford PhD, computational neuroscience. Google Brain, senior researcher, then team lead, then principal scientist. Publications in NeurIPS, ICML, ICLR — the top conferences, the ones where the real research appeared. Fifteen papers. Three patents. A Talent Resonance reading was impossible from a photograph, but the profile suggested someone who'd rate an 8 or higher. A genuine competitor, not a Gavin Belson proxy.
"This changes things," Ethan said. He pulled a chair to the conference table — the door-table, still the centerpiece of the office, now flanked by real chairs that Sarah had ordered from an office supply catalog. "Sarah, Marcus. Strategy session."
They sat. Sarah had her notebook — the systems diagram book, still the same spiral-bound volume from her barista days, though its contents had evolved from theoretical sketches to implementation notes for technology worth millions. Marcus brought his laptop with the monitoring dashboard open.
"Hooli's playbook for competition is three moves," Ethan said. "Acquire, poach, or destroy. They tried to acquire Pied Piper through a licensing offer that was really a control grab. They poached Pied Piper employees with inflated salaries. And they used legal pressure — IP claims, patent disputes — to try to destroy what they couldn't buy."
Sarah looked at him. "How do you know their playbook that specifically?"
"Public knowledge. Hooli's approach to Pied Piper was covered in tech press." True enough — the show's events had been positioned as his timeline's actual events, and the general pattern of corporate aggression was a matter of public record in any tech ecosystem.
"Which moves apply to us?"
"All three. But the order is different. Pied Piper had a product Hooli wanted — compression. They tried acquisition first because the technology was directly useful. We're building something Hooli doesn't have yet — they can't acquire what they don't understand. So the first move will be intelligence gathering. They'll try to understand what we've built before deciding whether to buy it, copy it, or kill it."
"The LinkedIn request," Marcus said. "From the Hooli engineer. Back in April."
"One of many, probably. Vincent Mora's first job will be mapping the AI landscape. We're on his map now — after the HooliBot tweets, we're impossible to miss."
Sarah was writing in the notebook. Not systems diagrams — a matrix. Threats on one axis, responses on the other. The engineer's version of a strategy document.
"Poaching is the immediate risk," she said. "Our engineers are underpaid relative to market. Google offers two hundred K plus. Hooli can match that. If Mora targets our core team—"
"He'll target our core team."
"Then we need retention. Equity vesting acceleration. Salary adjustments. Non-compete clauses."
"Non-competes are unenforceable in California."
"Then equity. Make it too expensive to leave." Sarah drew a column in her notebook. "Marcus, James, the two new hires starting next week — what are their ratings?"
The question was directed at Ethan, and it carried a weight that went beyond the practical. Sarah was asking him to use the Talent Resonance — the ability she'd never named but had tracked through months of observation — as a retention prioritization tool. Protect the highest-rated engineers first. Let the lower-rated ones become acceptable losses.
"Marcus is core team. Protect him first. The new hires — I'll assess when they start."
"And me?"
"You're not going to Hooli."
"No. But Mora might try. A CTO with hands-on Transformer experience and no non-compete? I'd be the first person I'd recruit."
She was right. In the chess game that was about to begin, Sarah was the most valuable piece on the board — a 9.5 with deep implementation knowledge, direct access to the architecture, and the kind of practical systems genius that Hooli's resources couldn't replicate through headcount alone.
"Your equity is ten percent, non-dilutable through Series A," Ethan said. "After the Series A dilution, you hold approximately seven percent of a fifteen-million-dollar company. That's worth over a million on paper. Hooli would have to offer you two million in guaranteed compensation to make leaving rational."
"Money isn't why I'd leave." Sarah set the pen down. "If Mora offers me a real research lab — compute, data, a team of PhDs, publication support — that's harder to counter than a salary bump."
"Then I need to make sure what we're building is more interesting than anything Hooli can offer."
"That's always been the deal."
Marcus had been listening without speaking — his default mode during strategy discussions, absorbing the dynamic between the two founders with the quiet attention of someone who understood his role was implementation, not direction. Now he spoke.
"There's a fourth move. One you didn't list."
"What?"
"Innovation. They could just... build something better. Mora has Google's research heritage, Hooli's engineering army, and unlimited compute. What if he doesn't need to acquire, poach, or destroy us? What if he just outbuilds us?"
The question landed with the particular discomfort of a truth nobody wanted to hear. Ethan's competitive advantage was the architecture in his head — the supernatural blueprint that gave him a three-year lead on the rest of the field. But a three-year lead wasn't permanent. It was a runway, and runways had ends. If Mora was an 8 — and everything about his profile suggested he was — then given Hooli's resources, the lead could shrink to two years. One year. Months.
The architecture progression helped — each generation unlock kept Ethan ahead. But the gap between generations required successful implementation, and successful implementation required time, compute, and a team that wasn't poached, distracted, or destroyed by a corporate adversary with four thousand engineers and the personal vendetta of its CEO.
"We stay ahead by moving faster," Ethan said. "The GPT model finishes training this week. We ship a product next month. We build customers while Mora is still staffing his team. The advantage isn't just the architecture — it's the execution. By the time Hooli builds something competitive, we'll be a generation ahead."
"That assumes they don't have their own breakthrough."
"Breakthroughs don't come from headcount. They come from—" He stopped. From a supernatural ability that deposits architectural knowledge in your brain. "—from insight. And Mora's insight, however impressive, is starting from 2014's understanding of neural networks. We're starting from—"
"From wherever you start from," Sarah said. "Which you still won't explain."
"Which I still won't explain."
Sarah picked up her pen and returned to the matrix. The conversation's unresolvable core — the gap between what Ethan knew and what he could say — settled back into its permanent resting place: acknowledged, unaddressed, tolerated.
The office hummed around them. Twenty-three people — twenty-six by next week, thirty by month's end, forty by December if the hiring plan held. The engineering section was loud with keyboards and low conversation. The sales team — Diana and a new hire named Derek, both rated 6, both performing above their numbers — occupied a corner near the window, making calls. The QA team tested the GPT-1 API prototype Marcus had deployed to an internal staging server the previous Friday.
Somewhere in Mountain View, in an office that was probably larger than their entire building, Vincent Mora was assembling a team to build what Ethan had already built. The race was on — not between two founders, but between a startup's speed and a corporation's scale. Between one person's impossible knowledge and an organization's unlimited resources.
Ethan opened his laptop. The ChronoCloud dashboard showed the GPT-1 training run at hour one hundred and eighteen. Loss: 2.41. Approaching convergence. The decoder-only architecture was producing outputs that were tighter, more creative, and more versatile than anything the original Transformer had generated. The GPT path was validating itself with every epoch.
His phone buzzed. Monica.
Vincent Mora just followed Gardner Analytics on LinkedIn. Thought you should know.
The observation was mutual. Mora was watching. Ethan was watching. Two chess players studying the board, each calculating the other's moves, separated by thirty miles of Bay Area geography and the fundamental asymmetry of a startup's agility against a corporation's mass.
Ethan typed back: Noted. Let's schedule a strategy session this week. We need to be ready for whatever comes next.
Sarah glanced at his phone from across the door-table. "Monica?"
"Mora's following our LinkedIn."
"Of course he is." Sarah returned to the matrix. Added a new row at the bottom: RESPONSE TO OBSERVATION — increase operational security. Compartmentalize architecture details. Restrict ChronoCloud access to core team only.
The pen moved. The matrix filled. Below the office, Manny's lunch rush was starting — the particular rhythm of a sandwich shop at noon, slicers and registers and the murmur of orders, a baseline of ordinary commerce underneath the extraordinary work happening one floor above.
Ethan watched the loss curve descend. 2.39. Each hundredth of a point was a step closer to a model that would prove the generative path was right, that the GPT architecture was the future, that the choice he'd made at a whiteboard in a SoMa office — creation over understanding, revolution over revenue — was the choice that built the world.
The curve dropped to 2.38. Mora was watching. Gavin was plotting. The twelve-month milestone clock was ticking.
And the model was learning.
Author's Note / Promotion: Your Reviews and Power Stones are the best way to show support. They help me know what you're enjoying and bring in new readers! You don't have to. Get instant access to more content by supporting me on Patreon. I have three options so you can pick how far ahead you want to be: 🪙 Silver Tier ($6): Read 10 chapters ahead of the public site. 👑 Gold Tier ($9): Get 15-20 chapters ahead of the public site. 💎 Platinum Tier ($15): The ultimate experience. Get new chapters the second I finish them . No waiting for weekly drops, just pure, instant access. Your support helps me write more . 👉 Find it all at patreon.com/fanficwriter1