Building the Anlace-class Light Frigate took a long time, much longer than I anticipated. It pushed me all the way into the eighth day of my two-week cycle, with the final piece settling in at around six in the morning. I even skipped sleep that night to get it down, working nonstop for nearly twelve hours. When I finally finished, slotting in and attaching the ship's turreted, the ship's primary weapon, I barely finished recovering from the new information the completion process downloaded before I crawled to my bed.

  I awoke later that day, at one in the afternoon, and spent a long lunch, with a large cup of coffee, going over not only what I unlocked with the final ship, but also what I unlocked consistently throughout the week with the other ships I built.

  First, I was confident I was done building ships and ship-related things, save for expanding my understanding of the MAC weapon systems. My understanding of Slipspace and the Slipspace drives was at the absolute peak the UNSC achieved, at least in the time span I had access to, ensuring I could quickly design a drive for whatever-sized ship I wanted. Well, technically, there was a minimum size as well, but beyond that, I was set.

  I had also gained a significant amount of knowledge beyond slipspace and space travel. A considerable portion of the civilian and military tech tree was cleared and locked in, with a clear understanding of everything from advanced microwaves and refrigerators to elevators and crane systems. I now had a massive database for civilian infrastructure at my fingertips, now at a level five hundred years past this world's standard. Noah was going to have a field day updating the internal plans for our two skyscrapers.

  I already had some of that from working with the vaults from Fallout, but this was such an advancement that it was incredible. I was particularly happy with my gains in materials science, which, unfortunately, didn't come in time to be included in the tower's primary infrastructure but would definitely be incorporated into later developments.

  Working with the Anlace had also helped us resolve our issue with installing shields on our towers. The UNSC studied and unraveled the mysteries of energy shielding in a few forms, some taken from the, while others were cracked from ship shields and the shields that the Elites wore.

  Halo shields came in a few different varieties, but all essentially functioned in the same manner, they were just implemented in different ways depending on which strengths the user found important and which weaknesses they could deal with. The way shields function in that universe relied on its particularly advanced understanding of quantum particle manipulation.

  Essentially, two layers of quantum-manipulated particles are formed, either into a specific shape or around a person or object. Between those two layers is a layer of excited plasma particles. This sandwiched particle layer interacts with the world as a solid barrier, as long as it is contained between the two layers of the quantum field. This is the shield barrier.

  However, when force is applied to the plasma barrier, it pushes against the inner quantum field, which glows as it attempts to mitigate the force. It even glows a different color depending on the quantum layer's frequency. Eventually, if you apply too much force, the quantum field will fail, releasing the excited plasma. In order to keep the plasma, which does contain a certain amount of energy, from hitting the user of the shield, both quantum fields are designed to vanish at once, which results in a much safer release of the agitated plasma.

  Once the quantum field generators have had time to recover and the plasma injector has had time to reset, the shields can begin to reform. The two quantum layers are formed first, and then the plasma layer is injected, slowly raising the shield's strength.

  This process took a lot of energy, which was one reason why Spartans had a fusion reactor on their backs. The Covenant apparently had a massive surplus of energy, meaning they could easily keep up with the monstrous energy consumption required to power shields constantly.

  The UNSC, on the other hand, came up with a much better solution. Rather than having their shields always on, they had them attached to a sensor field that encapsulated the entire vessel. The sensor would go off and activate the shield system whenever something dangerous threatened to hit the ship. Even better, since most structures and ships had plenty of space for multiple injectors, the shields could be activated and injected in split seconds.

  On top of that, by having the shields work in smaller sections, parts of the ship's shields could be activated while the rest were still off, further reducing energy consumption.

  Not only was it a fantastic system, but it didn't have the massive weakness to EMP that the Fallout universe's shield tech did, and didn't use special black boxed materials that the Titanfall universe did. Even better, they weren't mutually exclusive, meaning that if I was ever in the mood, I could use both shielding types on one structure of spacecraft.

  To say I was excited by the development was a massive understatement.

  Once I was done with my early lunch, late breakfast, I made my way around the town, taking a walk and unwinding after the blistering rush through the last week. The town's landscape had changed a lot while I had been busy, including the completion of not only the and but also the, which served as the cover for our new teleportation hub. The first of was nearing completion as well, and I got a sneak peek at what the, though there would be several versions throughout the structure.

  The garage workshop was more or less obsolete now, with Noah and Samwise both just waiting for permission to clear the area out and knock it down. It was determined that the center area, which was where the garage was, would be a central park, which Noah would be using to test out the limits of his rising and lowering gardens.

  I was excited to see how it turned out. The view from the top of the meeting hall was very nice, and I could see it getting even better as we expanded further. Of course, we needed to start finding people to live here.

  And that was the start of my first project for the day, which started by getting familiar with the UNSC AI systems. If I was going to start searching for people to live and work with my company and my buildings, I would need a way to check on them thoroughly. I could create the same type of AI that I had been making so far, passing them the task. But I wanted an AI that could go deeper. I wanted one that I could reliably trust to cut through any firewall, any system, any secret file they needed. Beyond that, I finally wanted a solution to the big bad AI monsters I knew were hiding all over this world, and behind the.

  Basically, I wanted my own Cortana.

  From what I had already seen, skirting around the edges of programming branches, I was almost completely certain that a properly specialized could handle the decentralized, almost Fae-like AI that infested this world. Especially if I used everything I just learned about UNSC computers and built them the most powerful server farm ever constructed to work from, which thankfully, Samwise had already been working on.

  The problem was that Smart AIs, which were created by destructive scans of human brains, were prone to a dangerous process called. Symptoms of rampancy include insanity, delusions of grandeur, god complexes, and an utter disdain for "lesser species," which unfortunately included humans. The only way I knew how to stop rampancy was euthanasia, and I was not willing to create an intelligent being with a life span of seven years. One, because it was cruel, and two, because reaching that point turned them into Netwatch's greatest fear.

  The other option, the "," were actually a mixed bag. They lacked any and all sentence or sapience, and were essentially extremely advanced response programs like Siri or Alexa. However, their programming was sophisticated enough, and their hardware powerful enough, that their utility and power was actually beyond the standard AI from Titanfall. Because there was no sentience or sapience, as long as their data was routinely checked, they had no risk of going rogue, meaning they could be fully decentralized, or unattached to a central core like my AI children.

  This made them incredibly useful, but also limited in their flexibility. A sufficiently advanced model with enough power behind it might be able to tackle some of the smaller AI around the planet, but that was not what I needed.

  So, to start off on my AI trip, I found a nice shady patch under the cover of the garage, wanting to enjoy my workspace one last time before it was torn down, and booted up my programming system. It had been a long time since I had had to learn any new programming, as I had pushed myself pretty hard in that avenue while I was still in the Titanfall tree. But it seemed like Halo had a few new things to teach me.

  I plowed my way through nearly two dozen different versions of Dumb AI, making quick work of them since half the time they used the same kernel core system with new modules or accessory programming. By the time I reached the peak in Dumb AI tech, about four hours had passed, and I had pretty much confirmed my previous theory that they would be insufficient for the larger AI.

  They would be very useful in other areas, such as quietly monitoring buildings or orchestrating factories. They were a fundamentally different design than the real AI I was capable of making, like Samwise and his siblings, so it was handy to know how to make them.

  I was also pretty sure they were the key to solving rampancy, but I needed to make a proper Smart AI in order to know that for sure. Thankfully, I could make one of those without turning it on, gaining the knowledge without condemning a sapient being to a seven-year-long life. I did something similar when I was first dabbling with Titanfall AI, so I was relatively confident in the process, even if this one was going to involve a deep, ultimately destructive scan of a clone of Sable's brain.

  I asked her specifically when I realized that I would need to make one, specifically one involved with electronic warfare. I would have used my own brain, but there was evidence that suggested personalities and even some memories might carry through, and while I was capable of it, I wasn't really a fighter at heart. Sable, while not necessarily born the badass woman she was now, definitely had more of a fighting spirit than I did.

  Unfortunately, despite the cloned brain being ready and waiting in the bio lab, we couldn't use it because the process required very specific hardware, which I would obviously have to make first. I ended up spending the rest of the day preparing the process, as well as preparing the program necessary to accept the brain scans themselves. I actually had to modify the process slightly after getting credit for the original, of course, since the AI was usually awoken immediately, and I wanted to avoid that.

  I went to bed on the early side so I could wake up the following morning well rested and ready to take advantage of a full day of building. It would be the ninth day of my Halo tech tree, and I was starting to feel the pressure. I had already made incredible progress just from working on the starships, but I still had so much more I wanted.

  When I woke up the next morning, Sable was waiting for me, sitting outside my room in the courtyard. I almost missed her, as I planned to take my teleporter to the bio lab, which was where we were running the experiment, but thankfully, Murtaugh left me a note written in the teleporter's terminal.

  When I left my apartment, Sable stood immediately, looking oddly nervous, noticeably enough that I stopped in my tracks. She readily agreed to everything when I first asked, but I was beginning to wonder if she was having second thoughts.

  This book was originally published on Royal Road. Check it out there for the real experience.

  "What?" she asked, looking down and flattening her business suit. "Something wrong?"

  "You're nervous," I said, raising an eyebrow. "You know, you can say you don't want to do this. I can have Frank incinerate the clone brain, and I can see if Kaytlyn or Jackie would like to volunteer."

  "It's less nervous and more… It feels like a responsibility, I suppose," she explained, furrowing her brow slightly. "It's not that simple, you understand that, right?"

  "Of course I understand, we are creating life, from your mind, it's not a small thing," I agreed, both of us walking towards the teleporter hub connected to the courtyard. "I do suppose I'm a little acclimatized to the idea from creating the other AIs, though. Its-"

  "Is she going to be my daughter?" She asked suddenly, snagging my arm as I stepped up the teleport platform, stopping me halfway up.

  When I recovered from her cutting me off, I felt rather stupid for not seeing the connection before. I had understood the trepidation about creating life, but had somehow missed the concern about creating life from her.

  "No… I think it's more like she will be your sister," I explained, putting my hand on hers, stepping back down to stand at her level. "She will be born completely self-sufficient, no child form or anything even close."

  The allegory of her being her sister seemed to soothe her nerves, and after a moment, she chuckled lightly.

  "Sister, huh? Well, having a sibling that I actually get along with might be interesting."

  As we made our way to the lab, I answered a few more questions, mostly just going over what I had already explained, as we walked to the bio lab, where Frank and Samwise were waiting. The bio lab was set up with everything we needed, including the cloned brain and the scanning equipment.

  I quickly went around doing a final checklist, confirming everything was in place and that the staging was ready to go. When I was confident the systems were ready, I nodded to Frank and Sam.

  "Alright, let's get this started," I said, standing next to the cooler-sized tank where the brain was floating. "Samwise, please keep an eye on the reading, let me know if any of the sensors leave the green. Frank, help me get this locked in."

  Frank helped me remove the specimen, the cloned brain, moving it into a specialized container. This was just about the most challenging part, since, despite what movies and fiction might tell you, the human brain has a consistency not too dissimilar from jello, making moving it risky.

  Once the brain was in its shell, we carefully put it inside the scanner, which would scan every neuron in great detail, before using the neural pathways to create the Smart AI's "," which was essentially the core of UNSC-style AI, both smart and dumb. For Dumb AI, it was basically just a complicated search-and-implementation algorithm, with the ability to add data relevant to its focus. It was obviously considerably more complicated for the smart variety.

  As the scan began, we had the holoprojector set up so we could see the matrix's progress, as it was created from the scans, layer by layer. Thankfully, the scan process itself was hidden behind opaque plating, as it wasn't exactly clean. The matrix began to slowly grow from the bottom up, formatting the AI's systems, an open and flexible matrix based wholly on Sable's brain. This was only the core framework, of course, which would be sort of nestled into its extra layers, feeding power into its intuitive and creative systems. It would learn, adapt, make connections, and grow over time, just like a human.

  And, in a normal Smart AI, this would eventually kill it.

  As the AI grew, in all aspects, it would slowly fill its Riemann matrix with increasingly dense pathways, each one a thought, a lesson, a chunk of data, or a learned ability. Because the matrix was finite, eventually this process would fill the matrix, ultimately resulting in an overload. This would exhibit itself as extremely unstable and volatile behavior. As its central core, in some ways, the matrix served the same purpose as the core did for stable Titanfall AI, even if it wasn't nearly as restricting.

  It was also why I was pretty sure I knew how to fix it.

  You see, the AI I made, the ones from Titanfall, technically had a finite amount of space in their core as well. I could offload that by giving them extra storage and even upgrading their cores, which I planned to offer them soon, since the UNSC had some ridiculous computing systems. But they also had pre-built-in systems to prevent their version of rampancy.

  They forgot things.

  Not quite like how humans do it, since when we forget something, it's usually gone forever. Instead, they took important stuff, used it to learn, and then took the unimportant stuff and either cleared it outright or boiled it down to simple numbers and facts.

  For example, if Samwise sat down with the group as we ate dinner, and we did nothing but talk about the weather, he would likely "forget" ninety percent of the conversation by the end of the month. If he considered it important enough, he might store the number of servings everyone had, their nutritional value, and how many cups of Jackie's latest concoction we had. If he considered it all unimportant, he might take note of the event, label it a "normal meal," and shove it into a spreadsheet labeled "shared meals."

  Everything else would be slowly cleared from his system, losing minor details over time, the older the "memory" was. Titanfall AI even had the ability to forget or boil down skills they learned that weren't important to their focus, though I cranked that down pretty hard when I programmed them. I wanted them to have hobbies, after all.

  Now, Halo Smart AI could compile and compress its data, but it was severely limited in both how far and how much it could do so, simply by the nature of its matrix. However, hypothetically, the information could be deleted from outside the matrix. It would just be an incredibly complicated and difficult process, something a human would never be able to keep up with, since the amount of raw data coming through, as well as the sheer complexity of the systems, was just too much.

  It would take something like a specially designed Dumb AI, programmed to piggyback on the Smart AI system, acting almost like its subconscious as it goes through its memories, curating, and collating the information. This Dumb AI could then either leave the data alone or gently cut it from the system over time, freeing up room for more information and learning. This Dumb AI could do this while still leaving the AI's mind and creativity intact, giving them a significantly longer lifespan. It was far from infinite, but I could see them living for several centuries.

  The Dumb AI would also serve as a limiter, keeping the Smart AI from lashing out at people if its systems started to degrade.

  Technically, this would prevent the sort of golden hour of Smart AIs in which the time before rampancy sets in, they are at their absolute best, since their matrix is absolutely as dense and informed as they could possibly be, giving them a massive amount of flexibility. However, with their extended lifespan and the slow clearing of memory and data that no longer served a purpose, they would likely become much more skilled at tasks they performed frequently, making them even more potent in their specialized field.

  While I was already pretty sure the concept would work, when the Riemann Matrix was complete, and therefore the AI was completed, I received a wave of information that confirmed it. It would not be a simple process, the Smart AI was capable of a massive amount of data throughput, and the Dumb AI "subconscious" would need to keep up with that. But it would be worth it to have a stable, superior AI that was capable of not only defending our Night City infrastructure, some of which would need to be connected to the greater net, but also take down problematic AI across the globe, when it came time to make some expansionist moves.

  With the matrix prepared and ready, its outer shell prepared, and its infrastructure ready, I got to work building the last piece. First, I created a Dumb AI from scratch, making it specifically for its role, leaving out anything connected to giving it a "personality," or even any communication skills. It had no reason to talk and wasn't alive in any way, shape, or form. Then I began integrating the two together.

  The installation processes took two hours, and felt almost like brain surgery. It was a legitimately stressful process, and I would be glad to pass it on to a more precise system once I had created one. When everything was complete, I ejected the AI's from the editing software, carefully pulling it from the specially made socket.

  "Is that her?" Sable asked, looking up from her tablet, where she had been working, staying close as I programmed and edited. "You said her core would be pretty small…"

  "Looks can be deceiving, but yeah, this is her," I responded. "This chip is the most advanced portable data storage device a universe close to five hundred years in the future could make. There's a lot of room for data on here."

  "But you have an even bigger mainframe for her, right?"

  "I do, Samwise built a special room, just for her," I responded. "Come on, let's get her settled in her new home."

  Sable and I both made our way from the bio lab to a side room filled almost entirely with the highest-level servers and computer processing systems I could create. At the core were four quantum computers, which I borrowed from a slipspace drive, as that was where they were predominantly used in the UNSC. These were the heart of the system, and were supported by a gigantic server farm filled with crystal data storage and computer systems. I was absolutely positive it was the most powerful computing system on the planet, and most likely in the surrounding space too.

  From here, our new AI friend would be able to reach all of our other systems and interface with a considerable number of robots and more. Some systems were monitor-only, with their entangled photons not directly connected to their systems, but almost everything would be at least visible from here.

  At the heart of the room was a simple station, almost a plinth, at the center of the room. It was little more than a holoprojector, a combination of all my tech trees so far, working together to produce the best-quality holograms I could. Small projectors like these were being installed all over the vault, including the apartments of those who consented.

  "This… why do I feel like this room is almost as impressive as when you introduced the teleporters to me?" Sable asked as we approached the pedestal at the center of the room. "What is all this?"

  "Each of those outer server platforms has enough power to engage in an exabyte of data analysis every minute," I explained, pointing to one of the outer crystal server stacks, one of fifty around the room, before pointing to one of the four quantum computers. "Each of the quantum computers-"

  "Quantum computers?" Sable asked, her eyes wide. "... That explains the."

  "Yeah, they run on something called cubits, and each one of these four runs twenty-five thousand of them," I explained. "Just one of them could most likely run all of Arasaka's computing at once. And our AI has four of them."

  Once we reached the pillar, I carefully slid the data crystal into the appropriate slot, standing back as the holoprojector lit up, dancing through a start-up sequence before finally revealing the newest member of the team.

  "Hello," she greeted easily, with a smile and a slight bow. ", it's good to meet you both."

  "Hello, Mary," I said, returning her smile. "It's good to meet you as well. How do you feel?"

  "I feel fantastic," She said, turning around on her pedestal, as if examining the room around her. "You gave me so much room to work, I really feel like I can stretch out."

  She emphasized the word 'stretch,' and as she did, the room's colors shifted, the RGB lights that signaled the status of all the cores and systems blinked, like a wave of color across the room. When it returned to normal, Mary let out a sigh, her form shimmering slightly.

  "Do I have a task that I should get started on?" Mary asked.

  "You do," I said with a nod. "Your primary task, the one you're always going to have, is to watch the net and the surrounding system infrastructure for threats. Leave the physical security to Murtaugh, unless he requests help, but defend your new home and family. Your secondary task is to get to know your new home, your siblings, and hopefully your new friends. Ask Samwise or me if you need anything built to help with either of those tasks. You should already have some holoprojectors to start from, and more are being installed as we speak."

  "I see, thank you, I will start with an internal security sweep, before setting up an outer perimeter," She says with a smile. "Are you concerned about NetWatch interference?"

  "Them finding out about our ability to create AI would be extremely problematic, but not world-ending," I admitted. "And while I'm relatively sure you are more than capable of hiding your presence on the net, don't underestimate them. Sometimes, for enemies like them, it's the low-tech tricks that catch you up."

  Mary frowned at my comment, taking my words into consideration before she nodded.

  "Very well, I will keep your words in mind."

  "Great. Now, unfortunately, I need to go," I said with a frown. "I only have a few more days with this tech tree, and I have so much more I want to learn. That said, feel free to stop by and chat, you should have access to my labs and workshops.

  "Of course… Would it be alright if I talked with Sable?" she asked, glancing at the woman in a nervous, almost embarrassed way. "I figure we should get to know each other…"

  "You're both free to talk to whoever you want, security and safety permitting," I said with a smile, glancing at Sable to see her eyes wide, looking at the projection. "Sable?"

  "Oh! Yeah, I have a few hours before I'm due at a meeting," the golden-eyed woman responded with a nod. "I'd be happy to talk."

  I smiled, waving to them both as I made my way out of the computer room, leaving them to their own devices. Even as I was just exiting the room, I could already tell their nervousness was settling, the two falling into a casual conversation about Sable's work.