For 30 advance/early chapters : p atreon.com/AutumnXd
Reeves walked Ryan out of the drift lab and toward the elite labs wing.
The five labs had been organized into two work clusters based on their research mandates. Ryan had divided the work that way deliberately, putting groups with related deliverables in adjacent spaces so they could collaborate without daily oversight from him.
Reeves pushed open the door of the first lab.
"This is the connection-persistence and stability team."
The researchers inside were gathered around a central worktable, examining several specialized components arranged across its surface. They looked up as the door opened. Several of them recognized Reeves and nodded.
Then they noticed the second person in the doorway.
Ryan had visited this lab approximately zero times since the team had been recruited. The shock of his physical presence in their space was visible. A few of the researchers exchanged looks that contained equal parts surprise, embarrassment, and the nervous calculation of whether anyone was caught playing games on their workstation.
Six months. The labs had been operating without their boss's direct oversight since the day they were brought online.
That had been a calculated decision on Ryan's part. He'd assigned focused mandates with clear deliverables and trusted Reeves to keep the labs on track. The researchers had been free to work at their own pace, attend conferences, take extended lunches. Compared to their previous institutional environments, where principal investigators monitored every research move, the autonomy had felt like vacation.
The autonomy had also felt like risk. The labs all knew what they were working on. Crimson Typhoon was the kind of project that produced careers. If they slacked, they were forfeiting their place in scientific history. The combination of light supervision and high-stakes work had created its own discipline. Nobody in the labs wanted to be the team that wasted a unique opportunity.
Reeves saw the wide eyes around the room and grinned.
"Today our base director and I would like to see your latest work. Show off what you've got."
"Don't call me a director," Ryan said. "I don't actually have a title here."
The room laughed. They all knew who the real authority was, regardless of titles.
Ryan exchanged greetings with the team and walked over to the worktable.
"I heard you ran a test yesterday. How did it go?"
A researcher named Frank Cabrera stepped forward to handle the briefing. Tall, mid-thirties, the kind of confident-but-self-effacing posture that suggested he'd been the senior tech in the team before joining Prism Sciences.
"Yesterday's test was successful, but Scrapper was too small a platform to fully validate the device. We didn't see anything anomalous, but we also couldn't push the system hard enough to find its limits."
He picked up a long rectangular case from the table. The case had a transparent center panel, allowing the contents to be seen through it.
Inside the case was an elongated metallic device. It looked like a stylized human spinal column, fabricated from segmented metal sections, each section connected to its neighbors by articulated joints. Each segment carried a small clamp on its forward face.
"This is the spinal clamp," Cabrera said. "Formal name: Total Neural Transmission Plate. We call it the spinal clamp because that's what everyone calls it."
The clamp's purpose was specific. In a Jaeger-scale mech, the pilot needed to be physically locked into the cockpit position with high precision. The spinal clamp accomplished this while simultaneously serving as the primary neural signal conduit. It locked along the pilot's spine, fitted a sensor against each vertebra, and amplified the neural signals from the pilot's spinal column directly to the mech's control systems.
In small mechs like Scrapper, a sensor cap and chest harness were sufficient. The pilot's neural output was relatively low-volume because the mech's command vocabulary was limited.
In a Jaeger, the command vocabulary was orders of magnitude larger. The mech had thousands of articulating components, each requiring its own control signal, all needing to be processed simultaneously. The pilot's nervous system simply couldn't generate enough signal for that volume of data through scalp sensors alone. The spinal clamp solved this by interfacing directly with the central nervous system at the source, bypassing the noise and signal loss inherent in surface-level acquisition.
Without the spinal clamp, three pilots could not control a Jaeger. With it, they could.
"Pull up yesterday's data," Ryan said. "Let me see the test logs."
Cabrera led him to a workstation and opened the relevant files.
"We made minor modifications to Scrapper to support the spinal clamp interface. Scrapper's command vocabulary is small, so we couldn't actually stress-test the clamp's bandwidth. But we did confirm that it transmits clean signals without interfering with the existing neural connection. Baseline functionality is verified. The bandwidth ceiling remains unknown."
Ryan settled into the chair and began reading the logs. The pattern of his focus was familiar to anyone who'd worked with him: eyes scanning the screen at high speed, but his attention not exactly on the screen, occupying some larger conceptual space where the data became three-dimensional. A pen turned slowly between his fingers.
The room went quiet.
Reeves wandered over to the spinal clamp display case and examined the device with professional curiosity.
He'd seen many of Ryan's technologies in this category by now. Each one had the same character: deceptively simple-looking on the outside, terrifyingly sophisticated on the inside. The spinal clamp looked like a slightly fancy back brace. It was actually a precision neural interface that could decode signals from individual spinal segments at clinical resolution.
The lab researchers exchanged glances. Each of them knew exactly how much work it had taken to translate Ryan's specifications into a working device.
Two months of pure analysis to even understand what Ryan had given them. The technical concepts had been pitched at the assumption that the readers were already at the frontier of their field. The labs had spent eight weeks just learning the vocabulary required to interpret the design notes. Then they'd spent six more weeks building a working prototype with Patricia's manufacturing partners.
And the entire concept had originated, fully formed, in Ryan's head.
That was the part the labs had stopped pretending to find normal. Their team of two dozen senior researchers had needed three months to fully understand a design that Ryan had produced as a single person. The asymmetry was disorienting. Most of them had quietly concluded that Ryan was operating on a different cognitive plane and stopped trying to understand the gap. Better to simply implement the work and accept the role of skilled hands building a vision that wasn't theirs.
Ryan finished reading.
"This looks good. The bandwidth assumptions in the model match the observed transmission rates, within margin. There's nothing here that contradicts the intended operating envelope."
He looked up.
"For full validation, we need to test the device on a system that can actually use the bandwidth. Scrapper's ceiling is too low. We'll have a better test platform soon."
Cabrera asked the obvious question. "Does that mean we test on the firefighting mech next? It has even fewer neural-controlled subsystems than Scrapper, so I'm not sure that helps."
Ryan shook his head.
"Different platform. More appropriate scale. I'll loop you in when it's ready. For now, keep the device in a stable state and document any anomalies you observe during the holding period."
He didn't elaborate. The test arm wasn't yet public knowledge among the lab teams, and Ryan preferred to keep information compartmentalized until each team needed to know.
Cabrera nodded. He'd worked under classified research conditions before. He understood "I'll tell you when you need to know."
Ryan thanked the team and headed out with Reeves toward the second lab.
The OS and AI lab occupied the adjacent room. The reception was identical to the first lab: surprise at Ryan's physical presence, immediate composure recovery, and a brisk transition to demonstration.
The principal investigator, a woman named Clara Morales, took the lead on the briefing.
"OS work is at twenty percent completion. The full operating system architecture is significantly more complex than we initially scoped. We're working through it methodically but it'll be a longer timeline."
"That's expected. Don't rush it."
"AI work is much closer to completion. We're at near one hundred percent for core functionality. Ready for system test."
Morales walked Ryan through the AI's capabilities. The mech AI's role was supportive: monitoring system status, alerting pilots to operational anomalies, automating routine functions, and adapting mech parameters to match individual pilot characteristics.
When the mech took damage, the AI would automatically run diagnostic protocols and report damage extent. When the pilot was distracted, the AI would handle peripheral situational awareness. When two pilots in the drift had different control preferences, the AI would mediate parameter adjustments to harmonize their inputs.
None of these tasks were individually difficult. The complexity came from coordinating them across thousands of subsystems in real time, integrating with the pilot's neural inputs without introducing latency, and ensuring the AI's behavior aligned with the pilot's intentions rather than overriding them.
The AI was, fundamentally, a copilot rather than a controller. It made the mech easier to operate by handling cognitive overhead, freeing the pilots to focus on tactical decisions.
"Ready for ground test on the firefighting mech?" Ryan asked.
"Yes. We have a test plan ready to go. We can run validation as soon as Marsh's team is available."
"Coordinate with them directly. I want results within two weeks."
"Understood."
Ryan thanked the team again. He and Reeves left the lab and walked the corridor back toward the main building.
Two of the five labs were on track. Three more to check. And the test arm waiting in the Storm Bay, growing more real every day as the technical crew assembled it.
The pieces continued to converge.