Kevin

My dad didn’t have to die.

That sentence is the one I have been carrying through this entire book. It is the sentence that the Foreword promised, that the Prologue gestured toward, that every chapter since has been quietly working toward. It is also the sentence that has the most weight inside the smallest number of words, because the version of his death that happened was not the only version that was possible, and the version that might have been possible if he had been diagnosed seven years later instead of seven years earlier is the version that the rest of this chapter is going to argue for.

This is the chapter where I tell you what happened, and what could have been different, and what I want you to take from both.

October 6, 2019

My dad died on a Sunday.

The hospital where the protocol had been run was in Seattle. The trial team was a group of doctors who had agreed to take him on when the standard protocols had stopped working and the experimental protocols were the only options left. The team was made up of some of the best oncologists in the country at what they did. They believed in the protocol they were running. They believed they could save him. They almost did.

The protocol was Keytruda combined with Interleukin-2. Keytruda is an immunotherapy that takes the brakes off the patient’s own immune system so the immune system can recognize the cancer as foreign and attack it. Interleukin-2 is an older immunotherapy that has been used for renal cell carcinoma for decades. The combination is aggressive. It is also one of the few options for Stage IV renal cell carcinoma when the cancer has spread beyond what surgery can reach. My dad had Stage IV renal cell carcinoma. The trial team’s combination protocol was the path forward they could offer him, and he chose it, and Linda chose it with him, and I chose it with both of them, and we chose it because the alternative was not choosing anything and watching the cancer take him on its own timeline instead of ours.

Linda has the medical timeline of those weeks more clearly than I do, because she lived inside it the way only the spouse of a cancer patient can — and because she had, herself, been a cancer patient twice before, navigating treatment for two unrelated cancers across the two decades that preceded my father’s diagnosis. By the time Kevin was on the operating table in Seattle, Linda was a two-time survivor herself, and she knew the inside of the cancer-treatment system the way most people only ever know it from the outside. When I asked her to walk me through the specifics of Kevin’s timeline for this chapter, she did. He was scheduled for three or four treatments over the course of four months. The first one began on September sixteenth, when he went in for the infusions. On September seventeenth, he was admitted into the ICU. He was supposed to be inpatient for seven days. The trial team could not get his heart to regulate during the protocol, and the seven days became ten. On the tenth day, the team made the call that he was stable enough to go home and recover before the next round. They told the family they had run the algorithms and thought it was safe to send him home. He went home.

Read that sentence again. “We ran the algorithms and think it’s OK to send him home.” That is the verbatim phrase Linda remembers the team using. The algorithms were the best decision-support tooling oncology had in 2019. They were running on the best signal the team could feed them. They said it was safe. The team trusted them. We trusted the team. He went home. The going-home decision is the structural connection between the 2019 trial team and the argument the rest of this chapter is going to make about AI in medicine, and I want you to register the connection now so it lands when I unpack it. The 2019 algorithms were operating with the technology of 2019. They missed what was happening to his heart. Better algorithms might not have missed it. We are going to come back to this.

He came home. He was supposed to return to the hospital on Monday, October seventh, for the second round of seven days. Linda had photographs of him during the inpatient stay that I have looked at since — him in the hospital bed with VR glasses on September twenty-ninth, when the team was trying to use the technology to ease the time he was spending immobile. Those photographs are timestamped. After September twenty-ninth, Linda told me, was when he started going downhill. The Keytruda and the Interleukin attacked his skin first. Athletes feet. Then the rest of his body. Then his heart. The same immune system the protocol had unleashed on the cancer was attacking the muscle tissue of his heart, and the heart was no longer pumping enough blood to his brain. He started having what looked like seizures. Linda was the one who recognized that whatever was happening was outside the boundary of what home recovery could absorb. She made him go to the emergency room.

Torel and Liz were already at the hospital with Linda by the time my dad was on the table. They had driven over to be with her the moment she had called them. By the time the medical team was working on him, Torel was twenty-one years old, Liz was twenty, and the two of them were the load-bearing people in the room. Linda was barely able to stand. The medical team was focused on my dad. Someone had to be the calm one. Someone had to be the strength when my mom had none. Someone had to handle the calls that needed to be made. That someone was my son. Torel was our rock that morning. Liz was the steady alongside him. The two young people who would six years later become the parents of my dad’s great-grandson were already, at twenty and twenty-one, functioning as the family’s operational unit at the moment the family was at maximum risk of falling apart. The line was not waiting for Dallas to begin. The line was already operational on October sixth, 2019, in an emergency room in Seattle, with my son holding my mother upright while my father was being worked on by people who had run out of options.

My mother called me first. That was the call I described in the Prologue — the cardiac-crisis call from Las Vegas, the cash in the airport, the couple that gave us their seats, the early scramble to get any flight north. We were already in motion. We were headed to McCarran. Then Torel called.

Torel’s call was the call that flipped the car. We were on the Vegas highway when his phone call came in. Whatever he told me — the escalation, the change in my dad’s status, the news that the situation was worse than the first call had suggested — was substantial enough that Michael cut the wheel and we flipped a bitch on the highway. We were now committed in a way the first call had not yet committed us. The morning had a different shape after Torel’s call than it had after Linda’s. The first call was the crisis. The second call was the turn.

Then I was on the plane. I cried for six hours straight. Michael sat next to me. He said nothing because there was nothing to say. An hour and a half from landing in Seattle, my phone buzzed. The buzz was a text from Torel. The text said:

“I’m sorry mom, grandpa didn’t make it, he’s gone. I love you so much.”

That is the message my twenty-one-year-old son sent me from the Seattle hospital after my dad died. “I’m sorry mom” is the lead. “Grandpa didn’t make it, he’s gone” is the news. “I love you so much” is the close. Twenty-one words from my son. The structure of the message tells you who he was at twenty-one. He led with the apology. He delivered the news in the most direct possible language. He closed with love. He could have waited for me to land. He could have left it for somebody else to tell me. He did not wait. He did not leave it. He was the one who picked up his phone in the Seattle hospital and typed the words and sent them so that I would know before I walked off the plane. That is who Torel was on October sixth, 2019. That is who I had raised. That is who, six years later, became the father of Dallas Kevin Bagley-Slone.

I read the text. I went into complete shock. And underneath the shock, somehow, I knew. I do not have a more sophisticated way to describe what happened in my body in the seconds after the text arrived. The mind had not yet absorbed what the words said. The body had. The body had emptied out before the mind caught up. That is what shock is, when grief arrives in the form of words on a screen on a plane an hour and a half from landing in a city you cannot get to any faster than the plane is taking you.

I texted my mother back. The text I sent her was nine words. “I love you so much mom, I’m sorry.” That was the message I had inside me to send. Not grief. Not anger. Not the questions I would have for the trial team in the days that followed. The text I sent my mother was the same architecture my son had sent to me. Apology and love. He had written “I’m sorry mom, grandpa didn’t make it, he’s gone. I love you so much.” I wrote “I love you so much mom, I’m sorry.” The Bagley women and the Bagley son had texted each other across three thousand miles within the same hour, in the same structural form, with the same two emotions named in different orders. I have never felt so empty as I felt sending that text. The plane was an hour and twenty-eight minutes from landing. My father had been dead for less than an hour. My son was at the hospital with my mother. My mother was alone in a way she had not been alone in forty-seven years. I was alone on a plane next to a husband who could not say anything because there was nothing to say. The whole family was alone in different rooms in different cities at the same moment. The text I sent was the only thing I could do from where I was.

He did not make it to October seventh. He died on October sixth, the day before he was scheduled to go in for the second round.

The cancer had largely been driven back by the protocol. The other kidney was clear. The adrenals were clear. The cancer that had brought him into the trial six months earlier was, in the parts of his body where it had been most established, in retreat. The treatment was working. The treatment was also doing something else. The same immune system that the protocol had unleashed on the cancer was attacking the muscle tissue of his heart. The medical term for that is myocarditis. It is a known risk of the regimen. The trial team had been monitoring for it. By the time the monitoring caught the level of damage, the damage was past what could be reversed.

He died on a Sunday. The cancer that had brought him to the hospital was being defeated. The treatment that was defeating it had killed him.

The Cocoon

Four and a half months later, on February 29, 2020 — a leap-day Saturday — we held my father’s Celebration of Life.

It was a Hawaiian-birthday-style gathering. My father had loved Hawaii. He had loved the music, the storytelling, the way Hawaiian gatherings let people speak about a person they loved without the formal-funeral choreography most American memorial services impose. We did it the way he would have wanted. Love, stories, music, and the people who cared about him most. Linda planned it. Linda hosted it. Linda made sure that the man she had been married to for forty-seven years got the kind of send-off that honored who he had actually been rather than the kind that funeral homes default to. The day was, for the family, the formal closing of the four-and-a-half-month period of immediate-aftermath grief that had begun on October sixth, 2019. It was the moment we put the love and the loss into a shape we could carry forward.

The next day was March first, 2020. The world began shutting down. COVID was announced. The pandemic that would define the next two years arrived twenty-four hours after we had finished saying goodbye to my father.

Most accounts of the pandemic year focus on what was lost. The funerals that did not happen. The gatherings that went to Zoom. The widows and widowers who had to grieve alone because they could not assemble the people who would have carried them through the early months. The collision of public catastrophe and private grief was the structural reality for millions of people in those years, and the literature about that collision is now substantial.

My mother’s experience of those months was different. When I asked her, near the end of my writing of this chapter, what the first year after my father’s death actually felt like for her, she sent me back a message that I am putting on the page in her own words because no version I could write would land what she said as honestly as she said it. This is Linda, in May of 2026, looking back at the year that began on February 29, 2020:

“Kevin’s Celebration of Life was held on February 29, 2020 — a Hawaiian-birthday-style gathering full of love, stories, music, and the people who cared about him most. The very next day, COVID was announced and the world began shutting down. In an unexpected way, that pause gave me space to survive the earliest part of my grief privately. The masks, isolation, and quiet meant I did not have to explain my heartbreak, hold conversations, or pretend I was okay before I was ready. While the world was grieving in its own way, I was grieving Kevin, and that cocoon of silence gave me time to breathe, cry, and slowly begin learning how to live without him.”

That cocoon of silence is one of the most quietly remarkable single perspectives on the pandemic year that I have encountered, and I am grateful my mother put it into words for this book. Most readers will recognize, in her framing, a perspective they have not yet seen articulated. The pandemic is usually described as the thing that made grief harder. For Linda, in the specific circumstances of February 2020 — four and a half months after losing the man she had been married to for forty-seven years, on the day after the formal Celebration of Life closed the immediate-aftermath period — the pandemic was the cocoon she did not know she needed. The masks meant she did not have to perform okayness before she was okay. The isolation meant she did not have to hold conversations she was not yet ready to hold. The quiet meant she could grieve at her own pace rather than at the pace the social-performance overhead of being a widow would have demanded. The world was grieving in its own way. Linda was grieving Kevin. The two grievings happened in parallel, and for Linda, the parallel was the gift.

That year of cocoon — the year between February 2020 and roughly February 2021 — is the year my mother began learning how to live without my father. The world was slowly reopening by the time the second anniversary of his death arrived in October 2021. She was different by then. She had had the time the world inadvertently gave her. She had had the silence. She had had what she needed.

The Cancellation

Two months earlier, my parents were supposed to be on a cruise to the Galápagos.

The diagnosis arrived in mid-to-late July of 2019. Linda has the original hand-drawn sketch from the doctor that came out of the appointment — a diagram of my dad’s kidneys and adrenal glands and where the cancer was located, drawn in real time during the conversation that turned the rest of the year into a different year. Linda saved it. She told me, when she described it to me later, that she saved it because she thought it was pretty cool. That sentence is Linda. The fact that she could look at a doctor’s freehand drawing of where her husband’s cancer was and register, in the same breath as the diagnosis, that the drawing itself was kind of cool — that is the woman my mother is. The same woman who would later sit at his bedside for ten days in the ICU. The same woman who would later make him go to the emergency room when his heart stopped regulating. She is steady, she is observant, and she finds the things in the world that are worth saving even when the world is handing her the worst news of her life.

My parents kept the diagnosis between themselves for a window of time. They did not tell me right away. They waited until August. They wanted, I think, to absorb it themselves before they handed it to their daughter. The cruise was canceled. The plan that my parents had been building toward — the trip, the months after the trip, the years after the trip — was replaced by a different plan that none of us had asked for. The new plan was three or four immunotherapy cycles, scheduled across four months. He had only one before he died.

On September fourteenth, two days before he was scheduled to go in for the first round of infusions, the family had a poker night with the grandkids. Liz and Torel were there. Linda was there. Kevin was there, in his element, doing what he had been doing for as long as I had been alive — being the grandfather at the table, teaching, joking, holding court over the same kind of casual family gathering that had defined our weekends for decades. The poker night was not anything special. It was a Saturday. Two days before the hospital. A normal evening with the kids who would grow up to be the parents of the great-grandchild he never got to meet.

I am writing the poker night down because the chapter that does not include it would be a chapter about a sick man and a hospital and a treatment and a death. The chapter that includes it is a chapter about a man who had a normal Saturday with his grandkids two days before the protocol started, and that ordinariness is part of what was lost. He was not a patient first. He was a grandfather, a husband, a developer, a real estate agent, a TV personality, and a thousand other things he had been across seventy-one years of being himself. The cancer interrupted the man. The chapter has to honor the man before it can honor the interruption.

My dad was a developer. Most of the readers of this book do not know that, because he was not a public-facing developer in the way that visible technology personalities are. He worked at Maverick Microsystems on the MICR encoder, the technology that allowed banks to read the magnetic ink on the bottom of checks before optical character recognition was reliable enough to do the same job. He shipped multiple Apple II games in the early 1980s, when shipping a game on the Apple II meant writing every line of the code yourself, drawing every sprite yourself, distributing the game on physical floppy disks that you mailed to publishers and waited weeks to hear back about. He taught my cousin Chris to code. He taught me to think about systems. He was, in every meaningful sense, an early-era technologist whose career happened to land in retail and finance instead of in the visible-technology category that would have made his name recognizable.

My dad was also a real estate agent. After his technology career, he and Linda built and ran Special Agents Realty in the Pacific Northwest. He was good at it. He had the kind of presence in front of a camera that made the work translate to television, and he ended up on multiple HGTV and DIY home-renovation shows over the course of his second career. People in the trades-and-television audience still see him on TV today. Linda will sometimes get a text from a friend or a stranger who has just caught a rerun of one of his episodes — “I just saw Kevin on TV.” That kind of message is the kind of strange continuing presence the dead leave behind in the era of streaming reruns. He keeps showing up on people’s screens. He keeps being recognized. The man who shipped Apple II games on floppy disks in 1982 ended up being the man who taught television audiences how to evaluate a property forty years later. Both careers were the same career, run by the same person, with the same instincts about how to communicate technical work to people who needed to absorb it without being technologists themselves.

My dad was also a black belt in Akedo. He had been training in the martial art for the better part of his adult life by the time I was old enough to remember, and the discipline of the dojo was as much a part of his weekly rhythm as the engineering work or the photography. He brought me to the dojo as a child, the same way he brought me to the office and the dark room and the garage. The dojo was another room in the house of my upbringing where my father was already inside something serious that the rest of the world did not yet take seriously enough, and where he made sure his daughter saw what serious practice actually looked like — the bowing, the patience, the years of repetition that produced the calm of a man who had been hit in the face often enough to know he could survive it.

My dad had also been around boats his entire life. The story of how he learned, which I want you to register because it explains the rest of his comfort on the water, started when he was fourteen years old. His mother — my grandmother — made the decision that most mothers in 1962 would not have made. She allowed her teenage son to go out on a great big cargo ship for three months. The captain of that ship befriended him over the course of the voyage and let him navigate. My dad learned navigation from a working ship captain on a working cargo run when he was fourteen. That is where the rest of his life on the water came from. The years he and Linda would later spend living on a three-story paddlewheeler in Lake Union, the years of small-boat handling around the Pacific Northwest, the calm he had on the water that most people never develop — all of it traced back to the three months his mother had been brave enough to send him out on. His brother spent his entire career in boating professionally. My dad carried the knowledge his whole life as a competent, lifelong navigator who happened to spend his career building software.

My father was a serious person and also funny as hell. The two were not in tension. He could be locked into a piece of code or a navigation problem or a real estate negotiation with the focus of a man who treated the work as something that mattered, and he could turn around in the same minute and pull a prank on whoever was within reach. He was quite the prankster. The household I grew up in was not a household where the technologist-father was austere or distant. It was a household where the father who could explain extranets to me at breakfast was also the father who would set up an elaborate joke at dinner that took hours to understand. He was brilliant. He was loyal. He was loving. Those three words are the words I want on the page about him because they are the words his friends would have used about him, the words his clients would have used about him, the words his coworkers and his dojo partners and his television producers would have used about him, and the words my mother would use about him today if you asked her to describe in three words the man she had been married to for forty-seven years.

He was good at the work. He shipped real things. He believed in technology in the way that the first generation of personal-computer developers believed in technology — not as a job, but as a worldview. He carried that same worldview into real estate. He carried it onto the television sets where he taught families how to evaluate the homes they were thinking about buying. He carried it onto the dojo mat where he practiced for decades. He carried it onto the boats he handled across the Pacific Northwest waterways. He carried it across every part of his adult life.

That is who got the diagnosis. That is who canceled the cruise. That is who hosted the poker night two days before the first round of infusions began. That is who chose the experimental protocol because the standard protocols had stopped working and the experimental protocols were the only options left.

He chose the protocol because he believed in trying. He had been a person who believed in trying his entire life. The Apple II games he shipped were trying. The MICR encoder work was trying. Special Agents Realty was trying. The HGTV episodes were trying. The decades of building and learning and watching his daughters grow up were trying. The Galápagos cruise was trying. The experimental protocol was the last version of trying he ever did. It is also the version that did not work. Or rather — it worked. The cancer was being driven back. The treatment killed him while the treatment was working.

The Autopsy

The autopsy report came back in the weeks after.

The cancer had largely been cleared from the parts of his body where it had been most active. The other kidney showed no cancer. The adrenals showed no cancer. The metastatic sites that the protocol had been targeting had responded. The protocol had done what it was designed to do, on the timeline the trial team had hoped for, with the specific outcome that the protocol existed to produce.

The cause of death was myocarditis. The same immune system that the protocol had unleashed on the cancer had attacked the muscle tissue of his heart and damaged it past the point where the heart could continue functioning. The autopsy was not ambiguous about this. The cure had killed him. The cure had also killed the cancer. Both findings were on the same page of the same report, and the family had to absorb both findings at the same time, and there was no way to absorb either one without absorbing the other.

I want you to register what that means, because the structure of this kind of grief is different from the structure of most other kinds of grief. He did not die because the doctors gave up. He did not die because the cancer won. He did not die because there was nothing left to try. He died because the thing that was working too well in one direction was also working too well in another direction, and by the time the second direction was visible, the damage was past reversing. The grief from a death like that is not the grief of failure. It is the grief of a partial success that came at a price the family was not asked whether it would pay. The price was paid anyway. The success was real. The success was also not enough.

Linda wrote a letter in the days after he died. I am not going to put that letter on the page in this chapter — it is hers, not mine, and the version of it I have permission to share is held in the bonus material at the back of this book. What I will say is that the letter named what was true: the man we lost was the man who would have been most excited about the technology that was about to arrive. The man who shipped Apple II games on floppy disks would have been first in line to understand what AI in medicine could do. The man who taught Chris to code would have wanted to be alive for the next chapter of computing. The man who believed in trying would have believed in the trying that the next decade of medicine was about to undertake. He did not get to see it. The reason he did not get to see it is the reason this chapter exists.

What AI Could Have Done

I want to be careful with this section, because the honest version of the argument is more nuanced than the version some readers will want me to make.

The version some readers will want is the version where AI would have saved my dad. It is not that version. The technology in 2019 was not at the level it would need to be at to identify the myocarditis risk profile in time for that specific patient receiving that specific protocol on the specific timeline he was on. The trial team that treated him was operating with the best monitoring available in 2019, and the monitoring was not sufficient to catch the cardiac damage in time. AI as it existed in 2019 would not have changed that. I want to say so clearly because medical readers will read this chapter and any overclaim about what AI could have done in 2019 will rightly be dismissed as the kind of magical thinking that damages the credibility of the larger argument I am about to make.

The honest argument is the prospective one. The next patient like my dad does not have to die the way he died. The AI infrastructure that is now arriving in oncology research, in clinical trial design, in real-time biomarker analysis, in patient stratification, in dose modulation, in side-effect prediction — the technology being built between 2024 and 2030 is going to give the next patient like him substantially better odds. Not because the same Keytruda-and-Interleukin-2 protocol will be different. The same drugs may still be used. What will be different is the sophistication of how the drugs are administered, monitored, and adjusted as the treatment unfolds. AI-augmented dose modulation will allow trial teams to titrate immunotherapy regimens based on real-time biomarker signals, slowing the dosage when the autoimmune response is approaching the threshold where it starts attacking healthy tissue. AI-augmented patient stratification will allow trial teams to identify which patients are at higher cardiac risk before the protocol begins, and to adjust the protocol accordingly. AI-augmented side-effect monitoring will allow trial teams to catch the early signals of myocarditis weeks earlier than current monitoring catches them, when the damage is still reversible.

None of this saves my dad. He died in 2019, in Seattle, on a Sunday in October. The technology that might have saved him did not exist yet. What the technology might do for the next patient like him is the argument this chapter is making. The next patient like him does not have to die the way he died. The next patient like him might get to come home from the hospital instead of dying in it. The next patient like him might get to go on the cruise that was canceled. The next patient like him might get to teach his grandson to code, the way my dad taught Chris. The next patient like him might get the years he was supposed to have.

That is what AI in medicine means to me. It is not abstract. It is not theoretical. It is not a research program that I am rooting for from the outside. It is a specific argument about a specific category of patient who, if the technology arrives at the speed and the quality the rest of this book has been describing, will be more likely to survive their cancer than my dad was. The technology has to arrive responsibly. The technology has to arrive with appropriate safety oversight, with appropriate clinical validation, with appropriate human-in-the-loop review at every decision point where AI augmentation could affect a patient outcome. None of those constraints are optional. The technology has to arrive responsibly because the alternative is the technology arriving fast and wrong, with patients harmed by AI errors instead of helped by AI capabilities. The responsible path is the only path that produces the outcome we want, which is more patients like my dad surviving the protocols that were supposed to save them.

[MEDICAL REVIEW REQUIRED: This entire section requires review by a board-certified oncologist or clinical pharmacist with experience in renal cell carcinoma immunotherapy regimens before publication. The Keytruda+Interleukin-2 protocol claims, the myocarditis-from-treatment causal claims, and the prospective AI-in-medicine argument all need medical-credibility verification. Do not ship this chapter without that review.]

My Dad Didn’t Have To Die

I want to close this book with what I believe.

I believe that AI in the right hands will save lives. I believe it will create new jobs that do not yet exist, ways of working that we cannot yet imagine, opportunities to do good in the world at scales that the previous decade could not have produced. I believe AI will propel the people who learn to use it well into a future that is moving faster than any previous technological era has moved. I believe we will get there.

I also believe we will have a messy middle. The next five to ten years are going to be hard. The technology is going to ship before it is ready in places where readiness matters. People will be hurt by AI deployments that should not have been deployed. Industries will be disrupted in ways that destroy livelihoods that were supposed to be stable. Communities will be reorganized by economic forces that they did not consent to and cannot easily defend against. The middle is going to be messy. The fears that responsible operators have about AI are not paranoid. The fears are legitimate. The technology is powerful enough that the harms it can do are real, and we should treat the harms as real, and we should build the systems and the oversight and the regulatory frameworks that prevent the harms wherever we can.

And I believe that if good people lead with the right intentions, we can accelerate the good that AI does in the world while reducing the harm. We can solve problems that have plagued mankind for centuries because we have not yet had the tools to solve them. We can give sight to people who could not see. We can give experiences to people whose bodies could not previously support those experiences. We can explore space at a scale that previous generations could only dream about. We can extend human life and reduce human suffering and produce the kinds of breakthroughs that, for most of human history, would have been called miracles. The technology is capable of that. The question is whether we, the people who deploy it, are mature enough to direct it toward those outcomes instead of toward the outcomes that produce short-term profit at long-term cost.

This requires the most mature version of ourselves.

It requires collaboration. It requires real conversations between the technologists building the systems, the regulators governing them, the operators deploying them, and the people whose lives will be affected by them. Real conversations — not the political-conference camp dynamics that the trades industry has been creating, not the polarized social-media discourse that has made every other technological transition harder than it needed to be, not the closed-room corporate strategy sessions that exclude the people who will actually live with the consequences. Real conversations across all of those constituencies, conducted in the spirit of the engineering communities that built the open internet — share what works, contribute what you have learned, treat the broader good as the operational priority instead of the proprietary advantage.

It requires testing and being okay with imperfections as we learn. The technology cannot wait for perfect. The technology will not be perfect. The path forward is to ship the work that has the heaviest impact, gather the data on what works and what does not, and rebuild based on what the data shows us. Speed and data are advantages here too — not just for an agency like CI Web Group, but for the entire civilizational project of integrating AI into the systems that govern human life. We learn by doing. We do by trying. We try in good faith. We accept that some of what we try will not work. We pay attention to what does work, and we expand it. We pay attention to what does not work, and we change it. That is how the messy middle becomes the future we are hoping for.

My dad didn’t have to die.

He was the kind of person who would have been one of the good people leveraging AI to make other people’s lives better. He was the kind of person who would have been excited about the technology I am about to spend the next decade of my life building. He was the kind of person who would have understood what I was doing, would have asked the right questions about the architecture, would have wanted to know how Chris was implementing the data layer, would have wanted to see Braedn’s agentic-system designs, would have been first in line to read this book. He would have been one of the good people. The world is poorer for not having him in this chapter of the technology story. So am I. So is my mom. So is my son Torel. So is my grandson Dallas, who will not get to know him, who will only know about him through what the rest of us remember and what I am writing in this book.

My dad didn’t have to die. He died because the technology that might have saved him did not exist yet. He died because the people who built the technology that did exist could not project compounding rates of change accurately enough to know what would be possible seven years later. He died because the medical research community had not yet integrated the AI infrastructure that would have given his trial team the ability to titrate the protocol differently. He died in the messy middle of an earlier transition that we, all of us, are now in the middle of finishing. The next patient like him does not have to die that way. The next patient like him will benefit from the technology that the rest of this book has been arguing for. The arguments in this book are not abstract for me. The arguments are about him. The arguments are about the next person like him.

This book is for my grandson. It is also for my dad. It is also for the next patient like him, who will not know my name and will not know my dad’s name, and who will benefit from the work that the people building the next decade of AI infrastructure are doing right now, in 2026, in agencies and labs and clinics and operations centers around the world. The work has to be done. The work has to be done responsibly. The work has to be done with the most mature version of ourselves leading it, with collaboration and real conversations and the willingness to test and learn and be wrong and try again. The work has to be done because the alternative is more deaths like my dad’s, and we have the chance, right now, to make sure those deaths do not have to happen.

My dad didn’t have to die.

Make sure the next one doesn’t.