41 min read
TWO PATHS
EDWARD
Friday, February 13, 2026 – 10:00 AM – Large Conference Room, Riverton Health Systems
The quarterly review told a story that numbers rarely told. Just not the story Edward had hoped for.
Edward sat in the conference room with Rachel Torres, watching her present the pilot results to the expanded leadership team. The headline number was 8%. That was it. Eight percent faster across the organization. The pilot teams had done better — Claims Processing was up 40%, Provider Portal 30% — but averaged across all of engineering, the number was eight.
Eight percent. After a year of mapping, training, coaching, fighting for budget, surviving the St. Augustine crisis, and personally facilitating dozens of workshops.
The room’s reaction told the rest of the story.
“Eight percent,” said Linda Park, VP of Platform Engineering. Her tone was careful. “For all the disruption, all the workshops, all the time my leads spent in mapping sessions instead of shipping… eight percent.”
“The pilot teams are much higher,” Rachel said quickly. “Claims Processing alone is—”
“The pilot teams have Edward in the room,” Linda said. “My teams don’t. They tried the mapping on their own and it was a mess. Three different interpretations of the framework. Two teams mapped the same process and got completely different results. One team spent two weeks on a value stream map and then went right back to the old way because nobody could tell them what to do with it.”
Miguel Santos, one of the pilot team leads, shifted in his chair. “It’s not that simple. The approach works when—”
“When Edward is there. Yes. That’s the problem.”
Edward listened, saying nothing. She wasn’t wrong.
“And then there’s the medication tracking feature,” Rachel said, pivoting. “Shipped in June. Now in twenty-three hospitals. One of them reported catching a potential drug interaction in the first week. A patient who might have been discharged with a dangerous combination.”
The room was quiet. That mattered. That was real.
“That’s what this is about,” Edward said. “Not productivity metrics. Not AI adoption rates. Building software that helps people.”
But Linda’s question hung in the air. Eight percent. For all of that, eight percent.
Wednesday, February 18, 2026 – 2:00 PM – Conference Room C, Riverton Health Systems
Not every team wanted to be mapped.
Edward stood at the whiteboard in Conference Room C, facing the Infrastructure Operations team. Twelve engineers, all of them staring at him with expressions ranging from skeptical to hostile.
“Walk me through what happens when a change request comes in,” Edward said. “Not the documented process. The real process.”
Silence.
Tom Reinhardt, the team lead, crossed his arms. “The real process is the documented process. We follow the runbook.”
“The runbook says change requests take three days for review and approval. Your metrics show average completion time of eleven days. Where do the other eight days go?”
More silence. A few engineers shifted uncomfortably in their seats.
“Look,” Tom said finally. “We’ve been through this before. Leadership comes in, asks us to expose our inefficiencies, promises it’s not about headcount. Then six months later, half the team is gone.”
“That’s not what this is.”
“You say that. Every transformation leader says that.” Tom leaned forward. “We’ve been burned before. At my last company, at the company before that. ‘Trust the process’ always ends with layoffs.”
Edward set down the marker. He understood the fear. He’d seen it before, in different contexts, with different teams. The instinct to protect territory, to hide problems, to avoid becoming the next efficiency target.
“Let me tell you what I’m actually trying to do,” he said. “There’s a medication tracking feature that’s been stuck for eighteen months. Nurses need it. Patients need it. Every day it doesn’t ship, someone might die because of a drug interaction we could have caught.” He paused. “I’m not here to cut your headcount. I’m here to figure out why work gets stuck. Some of that stuck work is in Infrastructure Ops. Some of it is in Product. Some of it is in my own office. I need to see all of it.”
Tom crossed his arms. “And if what you see is that we’re understaffed? That we’re drowning in requests and can’t keep up?”
“Then I need to see that too. Because right now, I don’t know. I have theories. Everyone has theories. But nobody has actually mapped where work waits.” Edward picked up the marker again. “Give me two hours. Show me the real process. If what I find is that you need more resources, I’ll fight for more resources. If what I find is waste that nobody’s ever questioned, we’ll cut the waste. But I can’t help if I don’t see.”
The room was quiet. Then one of the junior engineers, a woman named Amy Koh, spoke up.
“The handoff from Platform Engineering takes four days on average. They say it takes one. But they don’t have the context we need, so we have to chase them for clarification.”
Tom shot her a look, but she kept going.
“And the approval queue for production changes. Three people have to sign off, but one of them is always traveling. So we wait. Sometimes for a week.”
“Keep going,” Edward said.
For the next two hours, the team talked. They drew on the whiteboard, argued with each other, corrected each other’s assumptions. By the end, Edward had a map of the Infrastructure Ops process that bore almost no resemblance to the documentation.
Eleven days for a change request. Four days of waiting for clarification. Three days waiting for approvals. Two days of rework because requirements changed mid-stream. The actual work took about a day and a half.
“This is what I needed,” Edward said. “This is what nobody else has been willing to show me.”
Tom Reinhardt stood up, his expression still guarded but softer now. “What happens next?”
“I fix what I can fix. The approval queue with the traveling executive. That’s governance bloat. I can cut that. The handoff from Platform Engineering. That’s a communication problem. We can solve that.” Edward paused. “The clarification delays. That’s a specification problem. We need Product to write better requirements. That one’s harder, but it’s fixable.”
“And the headcount?”
“You’re understaffed. I can see it in the numbers now. But I can also see that half your capacity is being wasted on waiting and rework. Fix the waste, and your capacity doubles without adding a single person.” Edward extended his hand. “Thank you for trusting me enough to show me the truth.”
Tom shook it. “Don’t make us regret it.”
“I won’t.”
MARCUS
Friday, February 20, 2026 – 7:30 PM – Private Jet, Over Manhattan
The private jet banked over Manhattan, giving Marcus a perfect view of the skyline at sunset.
He was returning from his third investor meeting of the week, traveling with Catherine on the pre-IPO circuit. Boston to Chicago to New York, with San Francisco scheduled for tomorrow. Catherine handled the business narrative; Marcus handled the technology questions. Presentations in glass-walled conference rooms, dinners with institutional investors who managed more money than most countries.
The story was working. Every meeting, Catherine delivered the business narrative, Marcus explained the technology moat, and the investors nodded.
His phone buzzed. Sarah.
How was New York?
Good. Wellington Partners is in for $50M. That’s our anchor institutional investor.
Proud of you.
Marcus smiled and settled back into his seat. Eleven months from ringing the bell at the NYSE.
He was winning. He’d already won.
Everything else was noise.
EDWARD
Tuesday, March 3, 2026 – 9:00 AM – Edward’s Office, Riverton Health Systems
The industry conference invitation arrived the following week.
Rachel brought it to Edward’s office. “Healthcare IT Innovation Summit. They want you to keynote. Forty-five minutes on ‘AI in the SDLC: Why Process Improvement Beats Tool Adoption.’“
“How did they hear about us?”
“Word of mouth. Someone mentioned Riverton’s results, it got repeated.” Rachel sat down. “If you speak about this approach, it’s not just Riverton anymore; it’s the entire healthcare IT sector.”
“That’s a lot of exposure.”
“That’s a lot of impact.” Rachel paused. “Unless you want to stay below the radar?”
Edward considered the question. They’d been deliberately quiet, building proof, accumulating evidence. But at some point, “below the radar” became “hiding.”
“Set it up. But I want to focus on process improvement, not tools.”
“That’s exactly why they want you. Everyone else is talking about tools. You’re talking about actually looking at the process.”
MARCUS
Tuesday, March 10, 2026 – 2:00 PM – Axiom Boardroom
The board meeting in December was a celebration disguised as a meeting.
Diana Reeves presented the financials with barely suppressed excitement. Revenue up 12% quarter-over-quarter. Customer retention at 97%.
“IPO pricing discussions start next month,” Diana said. “We’re targeting $16 per share. $950 million valuation. Marcus, your stake would be worth approximately $13 million.”
The number landed in Marcus’s chest like a small explosion. Thirteen million dollars. Four years of sacrifice would pay off.
Victoria Hartwell smiled, a rare expression for her. “Congratulations are in order. You’ve built something remarkable.”
Afterward, Prakash texted: Can we talk? Something you should see.
Tuesday, March 10, 2026 – 4:30 PM – Marcus’s Office
Prakash arrived with a manila folder. He looked older than Marcus remembered.
“Three more customers this quarter mentioned private demos under NDA. They’re calling it ‘Project Prometheus.’ The capabilities they’re describing match our roadmap. Feature for feature. Plus things we haven’t even scoped.”
“Private demos aren’t a product. They’re vaporware.”
“I found a job posting. Anonymous company, ‘AI-native logistics platform.’ VP of Customer Success to manage ‘large-scale customer transitions.’” Prakash pulled another document. “Shell companies in Delaware. Engineers from legacy freight companies quietly going dark. Someone is building something big.”
Marcus opened the documents. His instinct was to dismiss them.
“Let’s say you’re right. So what? Even if someone ships something, it’ll be years before they threaten us.”
“What if Project Prometheus isn’t a startup? What if it’s someone with deep pockets and existing customer relationships?”
“Then we deal with it after the IPO.”
“After the IPO might be too late.” Prakash leaned forward. “The case studies talk about 4x, 5x productivity gains with AI. What if whoever’s behind this is experiencing those gains right now?”
Through the window, a container ship was passing. probably carrying a billion dollars in goods,
“Keep watching,” Marcus said. “Document what you find. After the IPO, we’ll address it properly.”
Prakash took the folder. His expression said he wanted to argue further.
“One more thing. Derek Huang. He’s been interviewing.” Prakash paused. “He mentioned ‘finding something worth building.’ And Marcus… he seemed excited. Like he knows something we don’t.”
EDWARD
Wednesday, April 15, 2026 – 2:00 PM – San Diego Convention Center
Edward stood backstage at the San Diego Convention Center, watching several hundred healthcare IT executives settle into their seats.
Jennifer had flown out with him. She was in the front row, giving him an encouraging thumbs-up.
He walked on stage.
“Good afternoon. I’m Edward Johnson, CTO of Riverton Health Systems. I’m here to tell you about a transformation we’ve been through, and why most AI transformations fail.”
He clicked to his first slide: What’s stopping you from shipping?
“We had a medication tracking feature stuck for eighteen months. So we mapped the value stream. Eleven weeks from requirement to deployment, but only two weeks of actual work. Nine weeks of waiting.”
He spent thirty minutes walking through case studies. Kevin Nakamura’s transformation. The Documentation team that eliminated redundant processes without any new tools.
“Start with where work waits. Not with what tools to buy.”
The applause was enthusiastic.
Afterward, Edward was surrounded by executives asking questions, exchanging business cards.
A CTO from a major hospital system: “We’ve been wrestling with AI adoption for two years. Now I think I know why: we’ve been buying tools instead of looking at our process.”
Rachel found him at the hotel bar.
“You’re going to need more business cards,” she said.
“I’m going to need more staff.” Edward ordered a bourbon. “This is bigger than I expected.”
MARCUS
Saturday, April 4, 2026 – 8:00 PM – Webb Family Home, Fort Lauderdale
The weekend conversation with Sarah happened over quiet dinner. The boys were at a tournament in Orlando.
“What are you hearing about Project Prometheus?”
Marcus set down his wine glass. “Rumors. Same ones I told you before.”
“It’s more than rumors now. One of the Garlands’ grantees heard a shell company called MAI Research LLC is approaching companies. Free migration from Axiom, free training, free support.” Sarah’s expression was carefully neutral. “They’re specifically targeting Axiom customers.”
“It’s a competitive tactic. Some startup trying to create uncertainty before our IPO.”
“‘After the IPO’ isn’t a strategy.” Sarah pulled her hand back. “Every time I raise concerns, you defer. What happens if ‘after’ is too late?”
“It won’t be too late.”
“How do you know?”
“Because I know this industry. I know our competitors.”
Sarah set her fork down.
“I hope you’re right,” she said finally.
She didn’t smile.
EDWARD
Sunday, April 12, 2026 – 7:00 PM – The Dock, Edward’s House on the Intracoastal
Sophia had spent the previous summer at Axiom. Marcus had offered her the internship back in February, a favor to family, a chance for Sophia to see “how a real company works.” Edward had been grateful. Sophia had been excited.
Now, six months later, she was home for winter break, and she looked troubled.
“How was the internship?” Edward asked.
“It was… educational.”
“That doesn’t sound like a ringing endorsement.”
Sophia was quiet for a moment. “Dad, can I ask you something? And you have to promise not to tell Uncle Marcus I said anything.”
Edward set down his fishing pole. “Of course.”
“Is it normal for a software company to not let engineers use AI tools? Like, at all?”
“What do you mean?”
“Catherine, the CEO, she sent out this memo before I started. Anyone caught using unauthorized AI tools would be terminated and lose their unvested equity. The engineers weren’t even allowed to experiment.” Sophia shook her head. “At school, we use AI for everything. Debugging. Code generation. Documentation. But at Axiom, everyone was typing everything by hand. It felt like going back in time.”
Edward felt a chill. He’d heard Marcus mention the policy, but hearing it from Sophia made it feel different. More real.
“Did you talk to Marcus about it?”
“I tried. Once.” She picked at a splinter on the dock railing. “He said the same thing he always says. ‘After the IPO.’ He said the board didn’t want any variables, and AI tools were a variable they couldn’t control.”
“What did the engineers think?”
“The ones who would talk to me? They were frustrated. A few of them were using AI on personal projects, just to stay current, but they couldn’t touch it at work.” She paused. “And I overheard something I probably wasn’t supposed to hear. Two senior engineers in the break room, complaining about the VP of Engineering. Derek something. They said he’d been interviewing, that he wasn’t even trying to hide it anymore. One of them said, ‘If Derek leaves, I’m right behind him. I don’t want to wake up in two years with obsolete skills.’“
Derek Huang. Edward knew the name. One of Marcus’s original hires.
“But here’s what really got me,” Sophia continued. “Things that would take me an afternoon at school? They took a week there. Maybe more. Not because the work was harder. Because of all the handoffs. I’d finish something, send it to someone for review, wait two days, get feedback, fix it, send it to someone else, wait another day.” She shook her head. “The actual writing the code part was fast. Everything around it was slow. Is this what my career is going to be like? Waiting for approvals and sitting in meetings about meetings?”
“It doesn’t have to be.”
“But that’s how it was at Axiom. And they’re supposed to be the innovative startup.” She picked at another splinter. “Dad, I started this internship thinking I wanted to work at a startup. Be like Uncle Marcus. Build something from nothing. Now I’m reconsidering. Is this what startups are actually like? Moving slower than everyone else because they’re afraid of risk?”
“Not all of them. Some startups are exactly what you imagined. Fast, experimental, willing to try new things.” Edward chose his words carefully. “But some get cautious as they get closer to big milestones. An IPO is a huge milestone. Marcus is trying to protect what he’s built.”
“Protect it from what? From getting better?”
Edward didn’t have a good answer.
“The thing is,” Sophia continued, “Uncle Marcus isn’t stupid. He’s really smart. But he’s so focused on the IPO that he can’t see anything else. Every question I asked about technology, about AI, about competition, he had the same answer. ‘After the IPO.’ Like the IPO is going to solve everything.”
“I’ve noticed the same thing.”
“So why don’t you say something? You’re his best friend. He might listen to you.”
“I’ve tried. At spring break. A few times since. He has answers for everything. Meridian is dying. Their competitors are years behind. The IPO is the finish line.” Edward sighed. “He’s never mapped his own process. Never identified where work actually gets stuck. He’s protecting a trajectory instead of examining it.”
Sophia was quiet for a long time. The sun had dropped below the horizon, painting the sky in reds and purples.
“I’m worried about him, Dad. And Aunt Sarah. And MJ and Isaiah.” She pulled her knees up to her chest. “What happens if the IPO doesn’t go well?”
“I don’t know. I’ve never been through an IPO.” Edward put his arm around her. “But I’m worried too.”
“Should I say something? Next time we’re all together? About what I saw?”
“I think Marcus needs to figure some things out for himself. But if you want to share what you observed, that’s your choice.” Edward paused. “Just be gentle. He’s carrying a lot of weight right now.”
They sat in silence until the stars came out. Somewhere on the water, a boat’s running lights blinked in the darkness.
“Dad?”
“Yeah?”
“I’m glad I’m at Riverton’s competitor.” She smiled slightly. “I mean, not competitor. But you know what I mean. Healthcare instead of logistics. At least you’re looking at your process instead of just protecting it.”
Edward hugged her tighter. “I learned from you. At spring break. ‘You have to know what you want.’ Start with the end in mind. Then figure out what’s actually stopping you.”
“I was just talking about hackathons.”
“Sometimes that’s how the important lessons get delivered.”
Wednesday, April 22, 2026 – 3:00 PM – Engineering Floor, Riverton Health Systems
Lisa Martinez stopped Edward in the hallway. She was beaming.
“I shipped my first feature today.”
Edward smiled. “Tell me about it.”
“The medication dosage calculator. The one that adjusts recommendations based on patient weight and kidney function.” Lisa had been a QA engineer for eight years. Six months ago, when Edward had restructured QA into a platform team, she’d asked to transition to development. “I never thought I could do this. I spent my whole career telling developers what was wrong with their code. Turns out I learned more than I realized.”
“Your QA background is an advantage. You think about edge cases that developers miss.”
“That’s what my team lead said. The other developers keep asking me ‘what could go wrong here?’ before they even write the tests.” She laughed. “I went from being the person who found the bugs to being the person who prevents them.”
Edward thought about the reorganization. QA and Security had become platform teams six months ago. Some people had struggled with the change. A few had left. But Lisa was the third QA engineer who’d transitioned to development, and all three were thriving.
“How’s the security tooling working out?”
“Ben Adler’s team built something amazing. The static analysis catches maybe 70% of the security issues before code review even starts. The developers are actually learning to think about security because the tools give them instant feedback.” She paused. “Ben told me he’s happier than he’s been in years. His team used to be the department everyone complained about. Now they’re building things that make everyone’s life easier.”
“No one can ever make security people happy,” Edward said. “Except apparently when you let them build tools instead of being a gate.”
Lisa laughed. “That’s what he said you’d say.”
Thursday, April 23, 2026 – 2:00 PM – AI-SDLC Board Meeting
Edward stood at the whiteboard, marker in hand. The AI-SDLC Board was assembled: Sandra Williams, Maria Santos, Johnny Morrison, Rachel, and this week’s rotating engineer, Priya Sharma.
“Let me walk you through what the pilot teams have built,” Edward said. He listed the agents: scrum master, security review, QA, SRE. Each one doing routine work so humans could focus on judgment calls. In the pilot teams, it was working. Security issues that used to sit in a queue for a week got a PR within minutes. Lisa Martinez went from running regression tests to shipping the medication dosage calculator. The audit trail was cleaner than when humans did everything ad hoc.
“Within the pilot teams, the results are real,” Edward said. “Forty percent faster in Claims Processing. Thirty percent in Provider Portal.”
Johnny leaned forward. “And outside the pilot teams?”
The room got quiet. Johnny had seen the same quarterly numbers everyone else had.
Edward didn’t flinch. “Eight percent across the organization.”
“Eight percent.” Johnny let that sit. “We spent $400,000 on AI tooling. You restructured governance. You’ve had my full support for six months. And the org-wide number is eight percent.”
“The pilot teams prove the approach works. The org-wide number proves we can’t scale it the way we’ve been trying.” Edward set down the marker. “I’m not going to stand here and spin 8% into a victory. It’s not what I promised. It’s not what I expected.”
Sandra was doing mental math. “Walk me through the economics honestly.”
“We cut $300,000 in retention offsites. Spent $400,000 on tools. Net cost of $100,000.” Edward paused. “Attrition in the pilot teams dropped from 18% to 9%. Outside the pilots? Attrition is actually up. Fourteen percent to 19%. Some of that is people frustrated by the inconsistency. Some is teams that tried the approach without support and got burned.”
The room absorbed that.
“So we have islands of excellence surrounded by organizational chaos,” Sandra said.
“Yes. That’s an honest description.”
Maria Santos from Legal spoke up. “The compliance implications of inconsistent methodology across teams…”
“Are real. We have three different versions of the mapping framework in use. Two teams that modified the agent configurations without going through governance. And the hospitals we onboarded too quickly are reporting performance issues.” Edward sat down. “The approach works. The approach at scale, inside an existing organization, with twenty years of legacy process? That’s a different problem. And I don’t have the answer yet.”
Johnny was quiet for a moment. Then, unexpectedly: “I appreciate the honesty.”
Edward raised an eyebrow.
“Six months ago, you would have spun this. Found a way to make 8% sound like a win.” Johnny shook his head. “The fact that you’re standing here saying it’s not enough? That’s more credible than any slide deck.”
“The medication feature is in twenty-three hospitals,” Rachel said. “Twenty-three potential drug interactions flagged. That’s lives saved.”
“That’s real,” Johnny agreed. “But it’s not an organizational transformation. It’s a pilot that worked and an org that couldn’t absorb it.”
Edward nodded. “That’s exactly what it is.”
The meeting ended without applause. No one owed anyone an apology. The pilot teams were proof of concept. The organization was proof that proof of concept isn’t enough.
Friday, May 1, 2026 – 4:00 PM – Edward’s Office
Success was supposed to bring better problems. Instead, it brought chaos.
Rachel knocked on Edward’s door. She didn’t sit down. “We have a situation.”
“Which team?”
“All of them.” She leaned against the doorframe. “Infrastructure Ops tried to run their own value stream mapping session. Without a trained facilitator. They ended up in a three-hour argument about whose process was worse. Tom Reinhardt walked out. Two of his engineers updated their LinkedIn profiles that afternoon.”
Edward closed his laptop. “That’s one team.”
“It’s not one team. The Platform Engineering group took our mapping template and modified it. Now they have their own version that doesn’t match anyone else’s. When they hand off to App Dev, the handoff documentation doesn’t align. It’s actually slower than before we started.” Rachel sat down. “And the medication tracking feature expansion? The three hospitals we onboarded last month without proper stress testing? Two of them reported performance issues. Not dangerous. But the kind of thing that makes compliance nervous.”
“How many teams are trying the mapping on their own now?”
“Twelve. Maybe fourteen. We trained six. The rest just started doing it because they heard about the pilot results.” She handed him a printed email chain. “Linda Park’s team created a Slack channel called ‘value-stream-skeptics.’ Forty-seven members.”
Edward read the email chain. Screenshots of botched mapping sessions. Complaints about inconsistent methodology. One engineer’s post: ‘Another management framework that works great in the demo and falls apart in production.’
“The pilot teams are producing real results,” Rachel said. “Claims Processing is still up 40%. The medication feature is saving lives. But outside the pilots? It’s a mess. And the mess is making the pilot results look like a fluke.”
“What’s the aggregate number?”
“Eight percent. Across the whole org, eight percent improvement.” Rachel paused. “That’s real. In a 400-person engineering organization, 8% is meaningful. But it’s not the story we told the board. And it’s not the story they’re going to want to hear.”
Edward stared out the window. Eight percent. The pilot teams were proof the approach worked. But the org couldn’t absorb it. Not like this. Not one team at a time, with him personally in every room, facilitating every session, resolving every conflict.
“I’m the bottleneck,” he said.
“You’re the bottleneck. And the approach doesn’t work without you in the room. Not yet.” Rachel’s expression was both sympathetic and brutal. “You built something that requires Edward Johnson to be in twelve places at once. That’s not a transformation. That’s a dependency.”
“So what do we do?”
“I don’t know. I wrote a proposal for training facilitators. Certification programs. Self-service toolkits.” She handed it to him. “But honestly? I’m not sure that fixes it. The problem isn’t that we need more facilitators. The problem is that you can’t transform an organization sequentially. One team at a time. One mapping session at a time. One Edward at a time.”
Edward read the proposal. It was thorough. It was also a band-aid.
“What would it take to do this in parallel? All teams at once?”
Rachel shook her head. “A completely different organizational structure. You’d need to rebuild from scratch. Parallel teams, parallel governance, parallel everything. You can’t retrofit that onto a company that’s been running serially for twenty years.”
The words landed like a diagnosis.
“So what we built works,” Edward said. “It just can’t scale inside what already exists.”
“Not without tearing everything down and starting over. And Riverton isn’t going to let you tear everything down.” Rachel stood up. “The medication feature is real. The pilot results are real. Twenty-three hospitals are safer because of what we did. That matters. But the dream of transforming the whole org from the inside?” She paused. “I think we hit the ceiling.”
Edward sat with that for a long time after she left.
Monday, May 4, 2026 – 9:00 AM – Edward’s Office
The recruiters started calling in January.
Not headhunters. Not small companies. The calls were coming from the biggest names in technology: a major cloud provider, a Fortune 50 tech company, an enterprise software giant.
The pitches were similar: We’ve been watching the medication tracking story at Riverton. We need someone who understands AI transformation.
Edward listened politely and declined. He had a job. He had teams counting on him.
But the calls kept coming. And after the quarterly review, after Rachel’s diagnosis, after staring at that 8% number, he started listening differently.
One call stood out. Cascade Technologies. 150,000 employees. The VP of Engineering said something that stuck:
“We’re not looking for someone to transform our existing org. We’re building a parallel organization from scratch. New structure, new processes, new governance. The old org keeps running. The new one proves a different model. When it works, teams migrate voluntarily.”
A parallel organization. Built from zero. No legacy to retrofit. No twenty-year-old processes to fight.
Edward didn’t say yes. But for the first time, he didn’t say no.
MARCUS
Tuesday, May 19, 2026 – 3:30 PM – Marcus’s Office, Axiom
The Whitfield analyst call changed everything.
Marcus had been preparing for the worst. February had been rocky: more questions about Meridian, more whispers about Project Prometheus, more nervous glances from his engineering leads. He’d started to wonder if the doubters were right.
Then Allison Marsh from Whitfield Capital called.
“Marcus, I wanted to give you a heads-up before we publish.” Her voice was warm, collegial. “We’re upgrading Axiom to ‘Strong Buy’ with a $22 price target.”
Marcus nearly dropped his phone. “$22? That’s 37% above our IPO range.”
“The fundamentals are compelling. Your retention numbers are industry-leading. The logistics market is consolidating, and you’re positioned to be the acquirer, not the acquired.” She paused. “We’re not seeing the Meridian threat that some of your competitors are hyping. Their platform is dated. Their customer base is shrinking. Whatever they’re building internally, it’s not showing up in market share.”
After the call, Marcus sat alone in his office, staring out at the port.
I was right.
All the doubters. Prakash with his manila folders. Sarah with her warnings. Even Edward with his value stream questions. They’d all been worried about nothing.
Meridian was dying. The analysts confirmed it. The institutional investors confirmed it. The IPO was going to be a success, and everything Marcus had sacrificed would be worth it.
He texted Sarah: Whitfield upgraded us to Strong Buy. $22 price target. Told you we’d be okay.
Her reply came an hour later: That’s wonderful news. Proud of you.
That night, Marcus slept through until 6 AM. No 2 AM anxiety spirals. No obsessive checking of competitor news.
Thursday, May 28, 2026 – 10:00 AM – Executive Conference Room, Axiom
The renewal report came in nine days later.
Customer Success had flagged twelve accounts as “at risk for renewal.” Twelve. In a single quarter.
“What’s driving this?” Marcus asked, the Whitfield glow still lingering.
Amanda Wong, Derek’s replacement, pulled up the customer feedback.
“Two themes. First: Project Prometheus. Some customers got approached by MAI Research LLC. They’re asking if we’re going to get leapfrogged.”
“Whitfield just upgraded us to Strong Buy. They said Meridian isn’t a threat.”
Amanda shifted uncomfortably. “Analysts measure market share. Customers measure features.”
“And the second theme?”
Amanda hesitated. “Our roadmap. Or lack thereof.”
“We have a roadmap.”
“Features they asked for in 2024 are now scheduled for late 2026. One customer said, and I’m quoting, ‘You’re slower than the dinosaurs we left.’“
Marcus felt the Whitfield glow fade. “We’re still nimble.”
“Nobody’s churning. But they’re signing one-year deals instead of long-term contracts.” Amanda paused. “They keep asking about AI features. About Project Prometheus. We keep saying ‘after the IPO.’“
EDWARD
Monday, May 11, 2026 – 6:45 PM – The Dock, Edward’s House
The text exchange with Marcus happened on a quiet Monday evening.
Edward was watching the sunset when his phone buzzed.
Hey brother. How’s Miami treating you?
Good. Been busy. Some interesting developments at work.
Edward wanted to tell Marcus everything: the value stream mapping, the medication feature, the conference keynote. But also the 8%. The org chaos. The teams that tried and failed. The feeling that he’d built something real but couldn’t make the rest of the organization absorb it.
But he remembered Marcus dismissing Sophia’s stories. He remembered “after the IPO” and the wall of confidence that nothing seemed to penetrate.
Our AI pilots are showing results in the teams that adopted it. The medication feature shipped. Industry conferences are asking me to speak.
That’s great. Glad it’s working out.
It’s more complicated than that. The pilot teams are strong but the org-wide numbers are… humbling. We should talk sometime.
Definitely. After the IPO, let’s grab dinner.
Edward stared at the response. After the IPO. The same phrase Marcus had been using for almost a year.
He wanted to push harder. He wanted to say Marcus, the world is changing faster than your IPO timeline.
But you couldn’t help someone who wasn’t asking for help.
Sure, he typed. After the IPO.
He set down his phone and watched a boat’s running lights move across the darkening water.
THE DINNER
Thursday, February 26, 2026 – 3:00 PM – Webb Family Home, Fort Lauderdale
Marcus insisted on hosting dinner at the Webb house. He wanted everything perfect: fresh flowers on the table, the good wine open, steaks on the grill.
The Johnsons arrived at three. Edward climbed out first, stretching after the drive from Miami. Sophia was home from Georgia Tech, a sophomore now, more confident than she’d been a year ago.
Sarah met them at the door.
“You look tired,” Jennifer said.
“It’s been a month.” Sarah didn’t elaborate.
Thursday, February 26, 2026 – 4:00 PM – Living Room
Edward noticed Marcus immediately: the tension in his shoulders, the rigid posture, the way his easy confidence had calcified into rigidity. Marcus looked like a man carrying weight he couldn’t set down.
When dinner conversation turned to Sophia’s semester, the energy shifted.
“Tell us about your classes,” Sarah said.
Sophia set down her fork. “The thing my professor keeps emphasizing is clarity. You can’t build well if you don’t know what you’re building. The AI amplifies whatever you bring to it, good thinking or bad thinking.”
Marcus looked up sharply. “In the real world, it’s more complicated. You need execution. Discipline. The ability to protect what you’re building from distractions.”
Sophia’s expression flickered but she didn’t argue.
Thursday, February 26, 2026 – 6:30 PM – Kitchen
Jennifer and Sarah loaded the dishwasher in silence.
“He’s not listening,” Sarah finally said. “To anyone. Every time I bring up AI, every time someone mentions competition, every time anyone suggests maybe the world is changing faster than his timeline, he says ‘after the IPO.’“
“How long has it been?”
“Since spring break. Nearly a year of ‘after the IPO.’ It’s become a reflex. A way to defer every conversation he doesn’t want to have.”
Jennifer set down a plate. “Edward’s pilots are producing results. The medication tracking feature shipped. But the rest of the organization…” She trailed off. “He doesn’t talk about it much, but I can tell. Some teams are thriving. Others are a mess. He’s frustrated.”
“I know. I’ve been hearing the success stories through my network.” Sarah’s voice was bitter. “‘Did you see that CTO from Riverton? The process improvement guy?’ My cousin’s husband is becoming a thought leader while my husband refreshes the ARR dashboard before breakfast. Meanwhile the thought leader comes home and stares at spreadsheets trying to figure out why it only works when he’s in the room.”
Through the window, the husbands were walking toward the dock.
“Maybe tonight,” Jennifer said. “Maybe Edward can get through to him.”
Sarah didn’t look hopeful.
THE BREAKTHROUGH
EDWARD
Monday, February 23, 2026 – 9:00 AM – Engineering Floor, Riverton Health Systems
That Monday, Claude Opus 4.5 dropped.
Edward was in his office when Rachel burst through the door, laptop open, eyes wide.
“You need to see this.”
She set the laptop on his desk. A code review was open on the screen. Not an ordinary review. A complete architectural analysis of Riverton’s medication tracking module, with suggested optimizations, security improvements, and performance enhancements.
“How long did this take to generate?” Edward asked.
“Forty-seven seconds.”
Edward stared at the screen. The suggestions weren’t just syntactically correct. They reflected deep understanding of healthcare compliance, HIPAA requirements, the specific edge cases their domain experts had documented over decades. The model wasn’t just generating code. It was reasoning about the domain.
“Run it again,” he said. “Different module. Something complex.”
Rachel pulled up the patient scheduling system. The legacy component they’d been afraid to touch for three years.
Forty-two seconds later, they had a complete refactoring plan. Not a generic suggestion. A plan that accounted for their specific database schema, their API contracts, their deployment constraints.
“This changes everything,” Rachel said.
“Stop.” Edward’s voice was sharp. “What’s the business outcome?”
Rachel blinked. “What?”
“You just showed me a faster way to generate code. That’s a task improvement. What’s the business outcome? How does this reduce days to ship? How does it affect patient care? Where does it fit in the macro flow?”
It was the culture he’d built at Riverton. Nobody talked about individual task improvements in isolation. Every conversation started with the business outcome and worked backward. The AI-SDLC Board had drilled it into every team: if you can’t connect an improvement to a measurable business result, you haven’t finished thinking.
Rachel nodded, recalibrating. “The patient scheduling system has been a bottleneck for eighteen months. Hospitals complain about double-bookings, missed appointments, cascading delays. If we can refactor it safely, we reduce scheduling errors by an estimated 30%. That’s fewer missed treatments, fewer patient complaints, fewer support tickets.”
“And the macro flow?”
“Scheduling feeds into the medication tracking module. Better scheduling means patients are where they’re supposed to be when medications are administered. The two systems compound.”
“Now you’re speaking Riverton.” Edward stood and walked to the window. “The model is more powerful. But the discipline doesn’t change. We start with outcomes, not capabilities. We map the flow before we optimize the pieces. The AI amplifies what we already know, but only if we know what we’re trying to accomplish.”
“So the framework matters more, not less.”
“Exactly. A year ago, AI could help with pieces. Now it can help with systems. But only if you know what system you’re trying to build. Only if you’ve mapped where work gets stuck. Only if you have domain experts who can validate what it generates.”
He turned back to Rachel. “Call a meeting. Everyone on the AI-SDLC Board. I want to understand what this means for our roadmap. And I want every proposal framed as a business outcome, not a capability demo.”
Monday, February 23, 2026 – 2:00 PM – Conference Room B
The AI-SDLC Board gathered around the table. Edward, Rachel, Johnny Morrison, Linda Park, and Dr. Amira Patel from Compliance.
“I’m going to show you something,” Edward said. “And then we’re going to talk about what it means. Not what it can do. What it means for our business outcomes.”
He demonstrated the Opus 4.5 capabilities. The architectural analysis. The refactoring suggestions. The speed.
Johnny Morrison leaned forward, his skepticism visible. “This is impressive. But it’s also dangerous. We don’t have governance frameworks for this level of capability.”
“Wrong framing,” Edward said. “You’re talking about the tool. What’s the business outcome we’re trying to achieve?”
Johnny paused. It was the question everyone at Riverton had learned to expect. No one presented a capability without connecting it to a measurable result. No one proposed an improvement without mapping it to the macro flow.
“Fair point.” Johnny sat back. “The business outcome is faster delivery of patient-facing features. Fewer medication errors. Better scheduling. Reduced support burden.”
“And what’s stopping us from achieving those outcomes today?”
“Legacy systems we’re afraid to touch. Technical debt we’ve been accumulating. The patient scheduling module that’s been a bottleneck for eighteen months.”
“Now we’re having the right conversation.” Edward pulled up the board charter. “The AI-SDLC Board exists to reduce toil, reduce risk, and deliver better software outcomes. That mission doesn’t change because the tools got more powerful. The question is always the same: what are we trying to accomplish, and what’s stopping us?”
“So what’s different with this model?”
“The amplification factor.” Rachel took over. “Before, AI could help with individual tasks. Code generation, documentation, testing. Now it can help with system-level thinking. Architecture. Integration. Optimization. But only if we frame the problem correctly.”
“Which means starting with the macro flow,” Edward added. “The medication tracking module worked because we’d already mapped the value stream. We knew the business outcome: reduce medication errors. We knew what was stopping us: an eighteen-month backlog. The AI didn’t replace that understanding. It accelerated the solution. Same discipline applies here.”
Dr. Kapur spoke up. “What about compliance? If the AI is generating architectural suggestions, how do we validate they’re HIPAA-compliant?”
“The same way we always do. Human review. Domain expertise. The AI is a tool, not an authority.” Edward paused. “But here’s what changes: the review gets faster because the AI can flag potential issues before we even look. It’s not replacing the compliance check. It’s enhancing it.”
Johnny was quiet for a long moment. Then: “What do you propose?”
“Controlled expansion. We take three teams that have already been through value stream mapping. We give them access to Opus 4.5 with specific guardrails. We measure the results. If it works, we scale.”
“And if it doesn’t work?”
“Then we learn why and adjust. That’s what the 90-day proof points are for.”
The board voted. Four to one in favor. Johnny abstained rather than oppose.
After the meeting, Johnny caught Edward in the hallway.
“I’m not convinced this is safe,” he said.
“I know.”
“But I’m also not convinced you’re wrong.” Johnny’s expression was complicated. “The approach you’ve built, the value stream mapping, the board structure, it’s more rigorous than anything I’ve seen. If anyone can make this work, it’s this team.”
“That’s not a ringing endorsement.”
“It’s the best I can do.” Johnny extended his hand. “Don’t make me regret it.”
Edward shook it. “I won’t.”
Tuesday, February 24, 2026 – 6:00 PM – Edward’s Home Office
That night, Edward sat in his home office, staring at his laptop.
The Opus 4.5 results were still open. The architectural analysis. The optimization suggestions. The speed.
Jennifer appeared in the doorway. “You’ve been in here for two hours.”
“I’m trying to understand something.”
“What?”
“Why this feels different.” Edward turned to face her. “Every AI advancement for the past two years has been incremental. Better code generation. Faster documentation. Improved testing. But this…” He gestured at the screen. “This is a step change. The model isn’t just executing tasks. It’s reasoning about systems.”
“Is that good or bad?”
“Both.” Edward stood and walked to the window. “It’s good because it amplifies everything we’ve built. The value stream mapping. The domain expertise. The process improvements. Teams that have done the hard work of understanding their own processes can now move faster than ever.”
“And the bad part?”
“Teams that haven’t done that work will fall further behind. The gap between organizations that understand their processes and organizations that don’t is about to get much wider.”
Jennifer was quiet for a moment. “You’re thinking about Marcus.”
“I’m always thinking about Marcus.” Edward sighed. “He’s five months from an IPO, and he’s never mapped his value stream. He doesn’t know where work gets stuck. He’s protecting a trajectory instead of examining it.”
“Can you help him?”
“I’ve tried. At spring break. Several times since. Every time I bring it up, he says ‘after the IPO.’” Edward shook his head. “You can’t help someone who thinks they’ve already won.”
“What happens when he finds out he hasn’t?”
“I don’t know. But I think we’re about to find out.”
Wednesday, February 25, 2026 – 10:00 AM – Engineering Floor
That Wednesday, Edward gathered the pilot teams.
“I want to share something with you. Yesterday, the AI-SDLC Board approved a controlled expansion of our approach. Three teams. Opus 4.5 access. Specific guardrails.”
Murmurs rippled through the room.
“But before we talk about what the model can do, let’s talk about what we’re trying to accomplish.” Edward walked to the whiteboard. “What are the business outcomes we’re pursuing?”
A senior engineer raised her hand. “Reduce medication errors. That’s the north star.”
“Good. What else?”
“Improve patient scheduling accuracy. Reduce support ticket volume. Faster time-to-market for patient-facing features.”
“Those are business outcomes.” Edward wrote them on the board. “Now, where does work get stuck in achieving those outcomes?”
The team called out bottlenecks. Legacy systems. Compliance reviews. Integration testing. Code review queues.
“This is how we think at Riverton. Business outcomes first. Macro flow second. Individual tasks last.” Edward circled the bottlenecks. “The new model is more powerful than anything we’ve worked with. It can reason about systems, not just tasks. It can suggest architectural improvements, not just code fixes. But that makes the discipline more important, not less.”
“Why?” someone asked.
“Because it’s tempting to get excited about capabilities. To show off what the AI can do. To measure task improvements in isolation.” Edward’s voice was firm. “That’s not how we work. Every improvement connects to a business outcome. Every optimization maps to the macro flow. If you can’t explain how a change reduces days to ship or improves patient care, you haven’t finished thinking.”
He drew a simple diagram: a foundation labeled “Business Outcomes” with “Macro Flow” above it and “AI Capabilities” at the top.
“This is what we’ve built. The foundation is knowing what we’re trying to accomplish. The middle layer is understanding how work flows through our systems. The AI capabilities sit on top. They amplify the layers below.”
He drew another diagram: “AI Capabilities” floating with nothing beneath it.
“This is what happens when you adopt AI without the foundation. Impressive demos. Fast task completion. No connection to business results. One strong wind, a competitor who actually understands their process, and it all collapses.”
“What’s the strong wind?” Rachel asked.
“Reality. A customer who needs something you can’t deliver because you optimized tasks without understanding the flow.” Edward set down the marker. “Opus 4.5 is a gift. But only if we use it to accelerate outcomes, not to skip the hard work of knowing what outcomes we’re pursuing.”
Thursday, February 26, 2026 – 7:00 PM – Webb Family Dock
After dinner, Edward and Marcus walked out to the dock.
The evening air was warm, the Intracoastal still and dark. They’d stood here a hundred times over the years, at cookouts, after football games, during late-night conversations that ranged from business to family to the meaning of everything.
“Five months to the bell,” Marcus said.
“Are you ready?”
“Ready as I’ll ever be.” Marcus stared at the water. “Victoria keeps asking about AI strategy. Prakash keeps bringing up Project Prometheus. Even Sarah’s worried.” He shook his head. “But I know what I’m doing. We’ve already won. We just need to execute.”
“What if the surprise is coming whether you’re ready or not?”
Marcus turned. “What do you mean?”
“The approach I’ve been developing. Value stream mapping. Looking at where work actually gets stuck.” Edward paused. “The pilot teams are seeing 40% improvement. But the honest number across the whole org is 8%. The teams I’m personally in the room for are transforming. The rest are struggling.”
“So it works but it doesn’t scale.”
Edward blinked. Marcus had cut right to it.
“That’s… yeah. That’s the problem I’m trying to solve.”
“That’s great for healthcare IT. We’re a startup. We move fast already.”
“Have you mapped your value stream?”
“We don’t need to. We’re cloud-native.”
“That’s an architecture choice, not a process metric.” Edward stepped closer. “Marcus, do you know how long it actually takes from idea to production? Have you measured it?”
Marcus didn’t answer. He’d never measured it.
“What if five months is too late?”
“It won’t be.” Marcus’s voice hardened. “I know this industry. I know our competitors.”
“You know what you knew a year ago. The question is whether the world is still the same.”
They stood in silence.
“Five months,” Marcus said finally. “Then all of this becomes permanent. Something I can give my kids.”
“I hope you’re right.”
“I’m going to ring that bell. When I do, all of this, worries, rumors, won’t matter.”
“Come on,” Marcus said. “The kids want to play cards.”
They walked back toward the house.
Thursday, February 26, 2026 – 10:00 PM – Back Deck
After the Johnsons drove home, Sarah found Marcus on the deck, staring at his boat.
“Edward tried to help you. To tell you about what he’s been doing.”
“He’s mapping value streams while I’m running a company. Different games.”
“Maybe they’re the same game.”
“They’re not.” Marcus didn’t turn around. “Five months, Sarah. That’s what matters. Everything else is noise.”
Sarah stood in the doorway for a long moment.
“I love you,” she finally said.
“I love you too.”
She went inside.