The Restructure

The Morning After

December 1, 2022. Kauai. Six in the morning, local time.

I had not slept much. I had, instead, sat at the desk in the hotel room with a notepad I had asked the front desk to bring up, writing down everything I had seen on the lanai the night before. By the time the Pacific started getting light outside the window, I had filled most of the pad. The handwriting on it got messier as the night went on. The last few pages are not quite legible to me when I look at them now. I was writing as fast as my brain could deliver. I was not optimizing for legibility. I was optimizing for not losing anything.

By six AM I had a list of names.

The list was the agency. 320 people. The full roster of CI Web Group as it stood on December 1, 2022. Account managers. Strategists. Designers. Developers. SEO specialists. Content writers. Project managers. Operations staff. Sales. Customer service. The full operational footprint of an agency that, by that morning, had been a generalist trades-marketing operation for almost seventeen years. 320 people who, the day before, had been doing the work the agency had been doing since 2010. 320 people who, by the time I finished my second cup of coffee, were going to be doing different work, or no longer at the agency at all.

I had not yet told any of them.

I was not yet ready to tell any of them.

What I had, sitting in that Kauai hotel room with the Pacific going from gray to silver outside the window, was the recognition from the lanai and a list of 320 names. What I did not yet have was the framework for connecting one to the other. I had to figure out, in the days and weeks ahead, which of those 320 people were going to be able to make the move I was about to ask the entire agency to make, and which ones were not.

That is what this chapter is about.

It is the operationally hardest chapter in this book. I have written it three times. Each version was wrong in a different way. The first version was triumphant — the heroic CEO who saw the future and acted on it, and look at how clearly the math worked out. That was untrue. The second version was apologetic — the CEO who had to part ways with 282 people and was carrying the moral weight of that for the rest of her career. That was also untrue. What I am going to write here is the third version, which is the only true one: the restructure was a discipline I executed, in real time, without knowing if it was going to work, while honoring the people I had spent years building the agency with as fully as the discipline allowed me to.

Some of them came with me. Some of them did not.

Both of those facts are part of the story.

The Hypothesis

Before I tell you about the framework, I have to tell you about the hypothesis the framework was built on.

The morning after the lanai, sitting at the desk in the Kauai hotel room with the notepad in front of me, I was not yet thinking about the agency’s organizational chart. I was thinking about energy.

Specifically, I was thinking about the energy economics of artificial intelligence. AI uses energy. A lot of it. The compute required to run a frontier model dwarfs the compute required to serve a search-engine result, and the compute required to run an army of autonomous agents will dwarf the compute required to run any current consumer application. That cost is not theoretical. It is the actual operating constraint of the era we are entering. And the implication of that constraint, I realized that morning, was that AI was going to be less patient with inefficiency than Google had ever been.

Google’s crawlers, for the last two decades, have indexed the web more or less indiscriminately. The crawlers do not care, much, whether your website loads in six seconds or one. They will index it either way. They will rank it lower if it is slow, but they will still index it. The economics of Google’s indexing operation could absorb the inefficiency of the web at scale because the marginal cost of crawling one more bloated page was small. AI is different. The agentic systems that are about to read the web on behalf of consumers will not absorb inefficiency the same way. They will skip it. The economics will not allow them to do otherwise. The web pages that win the AI era will be the ones built efficiently — lean code, fast load times, structured content, machine-readable architecture. The web pages that lose the AI era will be the ones built on bloated platforms with plugin sprawl and a code-to-content ratio that AI agents will read and dismiss in milliseconds.

That insight, sitting at a hotel desk on December 1, 2022, was the hypothesis that organized everything that came after. AI uses energy. AI will be less patient with inefficiency. The agency that is going to win the next decade is the agency that builds for AI’s economics, not Google’s.

And the implication for CI Web Group, I realized in the same hour, was substantial.

The vast majority of websites in the trades industry, including most of the websites we had built for our clients over the previous decade, were running on WordPress. WordPress is the most-used content management system on the web. It is also one of the least efficient. The plugin architecture that gives WordPress its flexibility also gives it its bloat. The PHP-based rendering pipeline produces code-to-content ratios that AI agents are going to penalize. The constant security-patch cycle that WordPress requires is itself an indicator of the platform’s underlying fragility. WordPress was the right platform for the era CI Web Group had been operating in. It was not going to be the right platform for the era we were entering.

So the first major decision I made on December 1, 2022, sitting in that hotel room, was that the agency was going to need to leave WordPress.

I did not yet know what we were going to leave it for. I did not yet know what we were going to keep, what we were going to build, or what we were going to release. What I knew was that the platform underneath the agency’s client work was about to become a strategic liability if we did not act. The hypothesis was clear. The execution was going to take the next eighteen months.

And the execution, I realized in the same hour, was not going to be possible while the agency was still pouring most of its operating energy into paid search and social media management.

PPC and SMM — paid search advertising and social media management — were two of the largest service lines at CI Web Group as of November 30, 2022. We were good at both. Both were profitable. Both had teams built around them. Both had clients who had hired us specifically for them. By every conventional measure of agency health, both were thriving.

They were also the two service lines that were going to consume all the operational capacity I needed for the AI restructure.

This is where I want to be honest with you about something the standard agency-pivot narrative gets wrong. I did not exit PPC and SMM because I had been quietly planning to for years. I did not exit them because I had decided they were dying service lines. I exited them because I needed the capacity. I needed the people-hours, the operational bandwidth, the leadership attention, and the financial runway that were going to PPC and SMM, and I needed to redirect all of it into preparing the agency for SGE and the AI search era. The exit was a sacrifice made for the AI bet. It was not foresight about the demise of paid media.

What I did not realize, at the time, was that the redirection was going to do something I had not predicted.

As we pulled the operating capacity out of PPC and SMM and poured it into the AI infrastructure work, our clients’ traditional organic search results began to improve. Their rankings on Google got faster. Their conversions increased. The work we were doing to prepare them for an AI-driven search era was, as a side effect, making them stronger inside the existing Google-driven search era. We had not bet on that. We had bet on the future. The present came along for the ride.

I want you to register that paragraph carefully because it is the answer to a question every contractor reading this book is implicitly asking. The question is: what if I bet on AI search and AI search does not fully arrive? The answer is that the bet improves your legacy search performance as a side effect. There is no losing version of this bet. The work that prepares your business for the AI era also makes your business stronger right now in the era we are leaving. That is the only reason I can recommend the bet to you in good conscience. If the bet were one-directional — only payoff in a future that may or may not arrive — I would not recommend it. The bet is bidirectional. That makes it the right bet to make.

So in late 2022 and early 2023, I began the work.

The Build

I want to walk you through what we built, because the build is the part of the restructure story that the standard CEO-pivot narrative skips, and the build is where the actual operating decisions live.

The first major move was the platform transition. WordPress out. Webflow in. We evaluated several content management systems through 2023, looking for the one with the strongest AI-enablement roadmap. The fully custom frameworks were cost-prohibitive at the agency’s scale. We settled on Webflow. The decision was made for four specific reasons: Webflow had cut the code-to-content ratio by roughly half compared to WordPress; Webflow eliminated the plugin attack surface, which made our clients’ sites significantly more secure; Webflow’s rendering pipeline produced the kind of efficient, structured output that AI agents are going to read and prefer; and Webflow had a credible AI roadmap that no other platform we evaluated was matching. The decisive moment was Webflow’s acquisition of Intellimize, an AI-native personalization company, which confirmed that Webflow’s leadership understood where the platform layer of the web was about to go. We made the bet. We began transitioning client sites off WordPress and onto Webflow.

That move made CI Web Group the first agency in the trades industry to transition off WordPress.

I want you to register that. Most of the agencies competing with us in the trades are still on WordPress as I write this in 2026. Many of them will eventually follow us. By the time they do, the AI search era will already be well underway, and the contractors who waited for their agency partner to lead will have lost ground that cannot be recovered. The first-mover position on the platform layer was one of the most consequential operating decisions of the restructure, and it was made on the energy-efficiency hypothesis I started this chapter with.

The second major move was the storage layer. We implemented Min.io — high-performance, S3-compatible object storage — to give the agency control over its own data architecture rather than depending on third-party cloud services that were going to become AI-era cost centers. Min.io is the kind of infrastructure decision that contractor readers will not feel directly but that the technical team feels every day. It is the foundation underneath everything we have built since.

The third major move was the agentic workflow build.

Starting in early 2023, we began building autonomous agentic workflows in-house. We chose n8n and LangChain as our primary orchestration tools because both gave us the flexibility to build custom workflows that no off-the-shelf agent platform could match. We have, as I write this in 2026, built over a thousand workflows on those two stacks. A thousand. That is not a typo. We have studied, tested, deployed, and iterated on more than a thousand agentic workflows for the work the agency does and for the work our clients’ businesses need done. Most agencies in the trades industry have built none. A handful have built a few. We are over a thousand deep, and we are still building.

The pace of the build was constrained by something the rest of the technology industry has been wrestling with too. The technology was changing faster than we could build.

By the time we had built one version of an agent or a workflow, the underlying technology had been replaced or extended by something better. We were building agentic workflows when MCP — the Model Context Protocol — was released, and we had to pivot the entire stack to incorporate it because MCP was clearly going to become the industry standard for how agents talk to each other and to data sources. The pivot cost us months of rebuild work. It also positioned us, by the end of the rebuild, on the right side of the standard. The agencies that did not pivot to MCP are now building on architectures that are about to be obsolete. We are not.

That pace — build, get replaced, pivot, build again — has been the rhythm of the agency since the lanai. There is no version of this work where you build something once and let it run for five years. The model providers ship new capabilities every quarter. The orchestration tools ship new features every month. The competitive landscape changes every week. The agency that wins the AI decade is the agency that has internalized the rhythm of constant rebuild as the actual operating cadence, not as an interruption to a more stable cadence.

The fourth major move was Synapse.

Synapse is the agency’s data layer — a unified data lake with retrieval-augmented generation, cache-augmented generation, and vector search built in. Synapse is what gives our agentic workflows access to the contextual data they need to operate intelligently. It is also what gives our clients access to their own data in ways that make the AI work possible. Without a Synapse-style architecture underneath the agents, the agents cannot do the work the agency promises. Synapse is, in 2026, the technical foundation underneath every other piece of the AI stack at CI Web Group.

The fifth major move was the autonomous agent layer.

We built autonomous agents that work alongside our team, not instead of them. The distinction is the entire point. The agencies that have tried to replace humans with agents are producing work that contractor readers will recognize on sight — generic, formulaic, missing the human texture of an industry whose customers are paying tradespeople to walk into their homes. The agencies that have refused to use agents at all are falling behind in capacity, throughput, and price. The right configuration is humans and agents working together, with humans handling the judgment, the relationships, and the decisions that matter, and agents handling the work that should never have required a human in the first place. We built the agency around that configuration. The 38 people who came through the restructure are working alongside agents we built ourselves, on a stack we built ourselves, with data infrastructure we built ourselves. That is the operating reality of CI Web Group in 2026.

The sixth major move was HydraOS.

HydraOS is the orchestration layer that ties all of it together — the platform layer (Webflow), the storage layer (Min.io), the data layer (Synapse), the workflow layer (n8n and LangChain), and the agent layer (the autonomous agents we built). HydraOS is the operating system of CI Web Group’s AI work. We built it with five specific objectives, and I want to lay them out here because they will tell you what we believed the right shape of an AI-era agency stack was when we designed it.

First: resolve security. The trades industry handles substantial customer data — home addresses, payment information, scheduling history, sometimes images of customers’ homes. The legacy WordPress-based agency stack was a security disaster waiting to happen. HydraOS was designed from the ground up to resolve the security exposure of the legacy stack. Our clients are safer running on our infrastructure than they were on the WordPress-and-plugin architecture we left behind.

Second: reduce code and increase Core Web Vitals. The bloated code of the legacy stack produced slow page loads, poor user experience, and AI-unreadable structure. HydraOS produces lean code, fast pages, and content that AI agents can read and rank efficiently. Our clients’ Core Web Vitals scores have improved across the board since the transition. That improvement is itself part of why their traditional Google rankings have improved as a side effect of the AI work.

Third: create an AI-first environment. The legacy stack was built for the human-browsing era of the web. HydraOS is built for the agent-browsing era. The architecture assumes that AI agents will be the primary readers of our clients’ content within a few years and is structured accordingly. We are not retrofitting AI capability onto a human-era stack. We built the stack for AI from the beginning.

Fourth: allow our clients to leverage our AI. HydraOS exposes the agency’s AI capability to our clients through interfaces they can actually use. A contractor running a home services business does not have the technical capacity to build their own agentic workflows. They do not need to. Through HydraOS, they get access to the workflows and the agents we have already built, configured for their business, with the technical complexity abstracted away. The contractor sees results. The technical complexity sits behind the curtain where it belongs.

Fifth: pivot major changes faster as algorithm changes accelerated. The rate at which Google, the LLM providers, and the standards bodies are shipping changes is increasing every quarter. The agency that pivots in three months when the market has moved in three weeks is going to lose. HydraOS is built to pivot fast. The architecture allows us to push platform-level changes across our entire client base in days rather than weeks, which means our clients are responsive to algorithm changes at a speed no manual-WordPress agency can match.

The seventh move was the product release. We released Cortex AI and Pulse — two products that emerged from the agentic infrastructure work. Cortex is the AI-search optimization product. Pulse is the local-AI-search product. Both are running in production for our clients, both are products of the build I have been describing, and both are reasons the agency’s clients are now winning AI-era search results that the rest of the industry has not yet figured out how to compete for.

That is the technology side of the restructure. The build is real. The build is at scale. The build is unique in the trades industry as I write this. And the build is not finished.

We have to keep working on this.

That sentence is not rhetorical. It is the actual operating posture of the agency. The build is never going to be done. The technology is going to keep changing. The standards are going to keep evolving. The model providers are going to keep shipping. We are going to keep building, pivoting, releasing, and rebuilding. The day we declare the build finished is the day we start losing. The day we declare the build finished is the day the standard CEO-pivot narrative would say we won. We do not believe that day exists. We believe the work is the rhythm. We believe the rhythm is the strategy. We believe the operators who internalize that are the operators who are going to be in business when the rest of the industry is asking how they got so far ahead.

I wear a sweatshirt to industry events sometimes that says SEO IS DEAD on the front.

It is one of my favorite pieces of clothing, and I wear it specifically because it gets contractors’ attention in a way nothing else I could wear would. They see the sweatshirt. They look at me. They look back at the sweatshirt. They get angry, sometimes, or worried, sometimes, or curious, sometimes, and they come over to ask me what I mean. And I tell them. I tell them that SEO is dead, not because it is, but because it has fundamentally changed and evolved into something the industry has not yet learned to call by the right name. The contractor who finds his Google rankings declining in 2026 because his agency is still doing 2018 SEO is not seeing SEO die. He is seeing the wrong version of SEO die. The right version is alive and growing and is being executed by the agencies that built for the AI era. That is what the sweatshirt actually means. It is a conversation starter, and the conversation is the same one I am having with you in these pages.

Now — with the build underway — the question I had to answer was who was going to do the work.

The Three Questions

By the time the plane landed in Houston on December 2, 2022, I had a framework.

It was three questions, written on the inside cover of the notepad. The questions were the framework. I have used them, in some form, in every operating decision the agency has made since.

The first question was: what work do we have to STOP doing?

Not slow down on. Not deprioritize. Stop. The lanai had told me that the era we were entering was going to require a different kind of agency than the era we were leaving, and the only way to make the move was to be honest about what work was no longer going to be part of who we were. PPC and social media management were the answer. Both lines exited the agency over the next eighteen months — wound down honorably, with clients given enough notice to find new partners, with staff given the option to retrain into the new work or to part ways with severance and references. Both lines were gone by the middle of 2024. Two service lines that had represented roughly half the agency’s revenue, gone by my own choice, on a timeline I had set.

The second question was: what work do we have to START doing?

The answer, on December 2, 2022, was longer than I expected. The agency had to build, from scratch, a working understanding of large language models. We had to build internal tooling that let our team operate alongside AI rather than in competition with it. We had to develop frameworks for how to write content for an era when search engines were going to summarize content rather than ranking it. We had to develop the AAO framework — Agent Answer Optimization — which I had scribbled on the notepad in Kauai and which would, over the next three years, become the operating discipline for how we approached every contractor’s digital footprint. We had to build technology infrastructure that supported all of it, and we had to build it on a schedule that did not allow for the usual two-year enterprise software development cycle.

The third question was: who on the team can come with us?

This was the hardest question of the three, and it is the question this chapter is mostly about.

The answer was not obvious in December 2022. It became clearer over the following six months as the agency began to actually do the new work. Some people, when given the chance to retrain, leaned into it. They picked up the new tools. They asked for the new training. They volunteered for the experimental work. They were the people who had always been most curious, most willing to be uncomfortable, most willing to be wrong while learning. Some people, when given the same chance, did not lean in. They wanted to keep doing the work they had been hired to do. They were good at that work. They had built careers around that work. They did not want to spend the next three years becoming a different kind of professional than the one they had spent the last ten years becoming. That was their right. It was also incompatible with where the agency was going.

The third question was not, ultimately, a question I was answering for them. It was a question they were answering for themselves. My job was to make the choice as honest as I could make it, give them the information they needed to make it, and then honor whichever way they decided.

That is the framework. It is three questions. They are not complicated. The complication is in the execution.

The Conversations

Between January 2023 and the middle of 2024, I had the conversation with every single person on the team.

Not all at once. Not in a town hall. Not by email. The conversation was, in almost every case, a one-on-one, in person or on video, in which I told the person what was changing in the agency, what was being built, what was being wound down, and what role — if any — was going to exist for them in the new architecture. I told them as honestly as I could what I thought their best path forward was. Sometimes the best path was inside the new agency. Sometimes it was outside. Sometimes I did not yet know, and I told them I did not yet know, and we agreed to revisit the question after a few months of running the new work and seeing how the chemistry held up.

I want to tell you something about those conversations that I have not fully told anyone in the three years since.

They were the hardest meetings of my professional career.

The reason was not that the people on the other side of the table were angry, or that they pushed back, or that they made the meetings harder than they had to be. Most of them did not. Most of them were professional, generous, and clear-eyed about what was happening. The reason the meetings were hard was that I had built the agency, in significant part, around the people I was now telling that the next version of the agency was not going to need them in the role they had been hired into. Some of those people had been with me for ten years. Some had been with me for fifteen. Some had been with me for almost the entire history of CI Web Group. They had been at our offices. They had been at our company events. Their kids had played with my kid at company picnics. They were not, in any honest sense, just employees. They were the people who had built the company alongside me.

And the meetings I was having with them were the meetings where I was telling them, as gently as I could, that the company they had built with me was not the company we were going to be from here forward.

I did not soften it. I did not pretend the change was smaller than it was. I did not promise outcomes I could not deliver. I did not invent reassurance that did not exist. I told the truth about what I was seeing in the market, what I was committing the agency to, and what role — if any — I was offering them inside the new shape of the company. I gave them every piece of information they needed to make their own decision. I tried to make the conversation as much theirs as mine. I tried to honor the years they had given me by being honest with them now, instead of stringing them along inside an agency that was no longer going to value the work they had spent a decade becoming excellent at.

Some of them stayed. Some of them did not.

The ones who stayed retrained. They learned new tools. They learned new frameworks. They learned to work alongside AI rather than in competition with it. They learned to charge clients differently, to deliver work differently, to think about their own roles differently. Some of them are now among the most senior people at CI Web Group. Some of them have built whole new disciplines inside the agency that did not exist in 2022. The transition was not easy for any of them. They earned every step of it. I am, three and a half years later, more grateful for them than I have any way to express in these pages.

The ones who did not stay went on to other agencies, other industries, other careers. Most of them are still in touch with me. Some of them are clients now. Some of them are partners. Some of them, when their next employer turned out to be the wrong fit, called me to ask if there was a way back. In a few cases there was. In most cases there was not, because by the time they called, the role they had been thinking about coming back to had become a different role inside an agency that had moved further than they realized. But the relationships endured. The way you part with someone matters more than the parting itself. I tried to part with every one of them in a way I could still look at, six years later, without flinching. I have not flinched.

The conversations were the hardest part of the restructure. They were also the most important part.

What I Kept

By the middle of 2024, the restructure was effectively complete.

CI Web Group had gone from 320 people to 250, then 140 — and the right-sizing that continued as the machines came online brought us to the number the agency runs on as I write this. PPC and social media management were no longer service lines. The agency was operating as an AI-first organic and local search infrastructure partner for trades and home service contractors and the companies that serve them. The technology stack we had built — some of which I will walk you through in a later chapter on HydraOS, OnePath.AI, and the broader ecosystem — was supporting work that, eighteen months earlier, would not have been possible. The clients who had stayed with us through the transition were doing better than they had been doing before it. The clients who had left, mostly because their service lines had been wound down, had been handed off to partner agencies that could continue to serve them in the categories we no longer offered.

Thirty-eight people.

That is what the agency looked like when the dust settled. Thirty-eight operators. Two-thirds of them retrained from inside the previous version of the company. One-third of them new hires, brought in specifically for the AI-era work. Smaller. Sharper. Faster. More expensive per head. More productive per dollar. More aligned. More honest about what we were building and why.

And substantially more profitable.

I want to put a single number on what the restructure did, because the financial outcome is the part most CEOs leave out of the restructure narrative and it is the part the contractor reader needs in order to evaluate whether the bet I am about to recommend is the bet they should make. When ChatGPT launched on November 30, 2022, CI Web Group was operating at high seven figures in revenue with three hundred-plus people. That was the snapshot at the moment the lanai recognition happened. By 2026, after the restructure, the agency was operating at eight figures with 38 of the best. The revenue category moved up. The headcount moved down. The dollars-per-head metric improved by an order of magnitude. The operating leverage that the AI infrastructure produced — the agentic workflows, the autonomous agents, the orchestration layer underneath — was real, was measurable, and was the reason the smaller agency was producing more than the larger one had.

I want to be honest with you about the full picture, though, because the standard restructure-success narrative would stop at *we got more profitable with fewer people* and skip the part that contractor readers need to understand. The full picture is this: the agency in 2026 is supporting three different business models concurrently. Our legacy clients are still operating on WordPress. Our Webflow clients are running on the platform we transitioned to during the first wave. Our HydraOS clients are running on the new stack we built. We have one hundred and thirty projects in production right now. The first dozen are live. The remaining hundred-plus are in active migration. The agency will not see significantly better profitability until all eight hundred of our clients are on HydraOS and we are no longer carrying the operational weight of running three concurrent stacks at the same time. That is the honest financial picture. The eight-figure revenue is being held against the operational cost of supporting legacy, Webflow, and the new build all at once, while we migrate the entire client base across the three. The real margin expansion will arrive when the migration completes.

Building AI infrastructure while AI is evolving at the speed of light is freaking hard.

That is the sentence I want you to register before any of the operational-leverage credentialing lands. Most CEO memoirs sand off the *freaking hard* parts of the work to make the operator look smoother than they were. I am not going to do that here. The work has been hard. The work is hard right now. Every week that passes brings new model capabilities, new orchestration tools, new standards developments, new agentic-AI breakthroughs that we have to evaluate and absorb and sometimes pivot the entire stack to incorporate. The migration of eight hundred clients from three different stacks onto a unified HydraOS architecture is not going to be finished tonight or this quarter or possibly even this calendar year. We are doing the work as fast as the technology will let us. The technology keeps changing. The work keeps stretching. The team keeps shipping. The clients keep migrating. The agency keeps operating. All of it at the same time. All of it freaking hard.

I want to tell you how we operate inside the freaking hard, because the operating philosophy is the part that matters more than the technology stack we have built. CI Web Group has not figured the AI transition out. We are fully building and rebuilding in flight, with the construction never quite finished and the destination moving every quarter. What we have built is not a finished system. What we have built is operational discipline for living inside an unfinished system. The discipline has two parts. First, we ship the heaviest-lift work before it is finished, gather the data on whether it is working, and rebuild based on what the data tells us. Second, we prioritize what has the heaviest impact over what looks the prettiest. The work that ships rough but earns the data is more valuable than the work that polishes a lower-priority deliverable to a finished state. Speed and data are our advantage. Beautiful and finished are not.

Speed and data are our advantage. We pivot fast.

That operating philosophy is the reason CI Web Group looks the way it looks in 2026. It is also the reason we are not the right partner for every contractor or every client. You have to be a little crazy to work with us. The work ships imperfect. The systems get rebuilt while clients are using them. The agency is constantly running experiments to see what works, and the clients who partner with us are part of the experiments whether they signed up to be or not. Our best clients are operators who tolerate the chaos because they want the speed. They understand that the alternative — working with an agency that polishes everything before it ships — means receiving deliverables that are obsolete before they arrive. The speed-and-data-and-pivot model is not for everyone. If you are looking for slow, pretty, and perfect, we may not be the best fit. If you are looking for fast, imperfect, data-grounded, and constantly improving, we are exactly the fit you have been looking for and probably have not yet found anywhere else.

There is one more thing about how we operate that I want to register honestly, because the book has been confident about CI Web Group’s work across many chapters and the honesty deserves the same register the rest of the manuscript has been using. There is not a lot of documentation on what we are building. The work we are doing at CI Web Group is on the bleeding edge of what is possible in AI-first marketing infrastructure, and the consequence of being on the bleeding edge is that there is no one to copy. There is no one to ask. There is, frankly, no one to even collaborate with at the level the work requires, because the operators who would be the natural collaborators are either inside the frontier labs themselves or are doing different work that does not intersect with ours. The category does not yet have widely accepted definitions. The peer-review infrastructure that mature business categories provide does not yet exist. The trade-press analyst who could independently verify the operational claims I am making does not yet have the framework to verify them, because the framework itself is part of what we are building.

I believe Hydra OS is one of the most advanced AI-first web and marketing platforms ever built. I am writing that sentence in print. The claim will be tested by time. I am making it anyway, because I believe it is true and because the work I am describing across the rest of this book is built on top of it. The honest version of the claim includes the acknowledgment that I am the operator making it about my own platform, that the platform does not yet have the external validation a more mature category would provide, and that the readers who will eventually evaluate the claim include operators who are building competing platforms and who will have their own views. All of that is fair. I am making the claim anyway, because the alternative — staying quiet about what we have built and waiting for someone outside the company to write it up first — would mean not telling the truth about the work. The book is not the place to stay quiet about the work. The book is the place to tell the truth about it and let the truth be tested in public.

Here is the operational evidence underneath the claim. CI Web Group has, in 2026, deployed our first team of autonomous agents into the daily operations of the company. Each one was built and trained by a specific human inside the company who took ownership of the agent and its work. The agents have names. They are not generic AI tooling. They are named, individually-built, individually-trained autonomous systems running real operational work alongside the humans who built them.

Braedn Heney built and trained Godspeed. Allula Teka built and trained Groot. Michael built and trained Sarge. I built and trained BossBot. Chris Heney built and trained Wolverine — and Wolverine is, frankly, a bad ass. Each agent operates in the workflow domain its builder owns. The first wave was the senior technical team plus me and Michael, each of us taking responsibility for an agent that runs the work we know best. The second wave is now in production. My senior leadership team — Melanie Osio, Jazmin Hernandez, Chelsea Nodine, and Michael Parker — have each built their own autonomous agents driving major pieces of the company’s strategic execution. Most agencies have AI tooling that the engineering team built for the rest of the company to use. We have AI agents that the leadership team built for themselves to use, with each leader taking ownership of what their agent does and how it operates. That is a different organizational architecture than what most agencies operate with. Democratized agent-building across the leadership team is one of the structural reasons the company can move at the speed it moves at.

The infrastructure underneath the agents includes a system called Synapse, built by Chris Heney with Marcos, our developer who has been deep in the Hydra OS architecture for years. Synapse is the knowledge-graph layer that sits between our language models and our action systems — the part of the architecture that holds the structured representations of our clients, their businesses, their service areas, their products and pricing, the relationships between all of those entities, and the ways those entities map to the public web and to AI search. Synapse is one of the load-bearing pieces of Hydra OS, and Chris and Marcos are the humans whose work made it real.

There is also a transformation I want to register that does not get enough credit when people talk about the AI era — the transformation of the design function. Most readers will assume creative designers are the slowest profession to AI-transform because design is creative work and creative work is supposed to be where humans hold ground. The opposite has been true at CI Web Group. Michael Parker, our Executive Director of Design, has fully transitioned from traditional creative designer through multiple phases of AI — from AI-augmented wireframing, through AI-driven design frameworks, all the way to full vibe coding. He is, in 2026, building production interfaces by directing AI rather than by manually composing pixels in design software. That is the most aggressive AI-adoption arc a creative designer can have, and most senior design leaders have not yet completed it. Michael Parker has. The fact that the design function at CI Web Group has fully transitioned at the same depth as the engineering function tells you something about how the AI-first transformation here is structured. Engineering did not transform alone. Design transformed too. The transformation is company-wide.

There are also relationships I want to disclose because the book has been operating in the epistemic-honesty register and the disclosure architecture should match. CI Web Group does not just operate as a standalone agency. It operates inside an ecosystem of adjacent companies where I hold structural roles — board seats, small equity positions, sustained collaborative relationships with the founders — and where CI Web Group provides the operational delivery layer of marketing, websites, or both.

I sit on the board of OnePath.AI, a company in the trades-industry technology space founded by Utku Kaynar, with Lynn Wise as its primary investor. I hold a small equity position in OnePath.AI. CI Web Group does the marketing for OnePath.AI. Utku, Lynn, and I have a strategic collaborative relationship that goes beyond the marketing engagement — we are operating in the same ecosystem, on adjacent problems, with overlapping investor and operator networks.

I sit on the board of Trade Rated, a company founded by Paul Wiese. Paul holds the trademark on the tagline “Built By The Trades,” which is the brand identity Trade Rated operates under. The company has a portfolio of nine apps in production. I hold a small equity position in Trade Rated and its nine apps. CI Web Group provides the websites for Trade Rated. Paul and I together provide the strategic insights that shape the company’s direction.

Paul Wiese is also a CI Web Group client through the company he founded called Doorserve Pro, which was recently acquired by Guild Garage. Paul appears later in this book as one of the named operator voices in the coalition chapter, and the relationship architecture between Paul and me — founder of Doorserve Pro, founder of Trade Rated, holder of the Built By The Trades trademark, my client, my collaborator, my fellow board figure — is more layered than the coalition-chapter quote alone would suggest. I am disclosing that here because it is the kind of structural relationship most CEO memoirs would handle vaguely or not at all, and I would rather be direct about it. Paul is not just a like-minded peer. Paul is structurally inside my ecosystem at four distinct layers, and the book is more honest with the architecture on the page than with the architecture hidden.

That ecosystem of adjacent companies, board roles, equity positions, and operational delivery relationships is part of how CI Web Group has built what it has built. We are not just an agency. We are the operational engine inside a larger ecosystem of companies operating in the trades-industry technology space, with shared investors, shared operators, shared strategic problems, and shared progress. The book has been pointing at this without naming it directly. I am naming it now. The architecture is legible. The relationships are disclosed. The work continues.

The reorganization underneath all of it was deliberate. We rebuilt CI Web Group around four functional groups.

First, a tighter leadership team. The directors and senior operators who had earned their positions through the restructure, with clearer authority and faster decision cycles than the previous version of the company had supported. Second, an AI engineering team — the technical engineers who build, ship, and maintain the agency’s AI infrastructure across HydraOS, Synapse, the agentic workflow stack, and the autonomous agent layer. Third, a vibe code team — operators who work in the loop with AI systems to produce the kind of fast iterative builds that the AI era requires, with humans and agents collaborating in real time on the work that used to take weeks of traditional development. Fourth, a client-facing growth division — the operators who interface directly with our clients, supplemented by a few subject-matter experts for the technical marketing depth that some clients need at moments of high complexity. Four groups. Thirty-eight people. Eight-figure revenue. Eight hundred clients. One hundred and thirty projects in production. Three concurrent business models in transition. That is what the agency looks like in 2026.

The 38 who stayed are the agency now. They are the people who run the day-to-day. They are the people who interface with the clients. They are the people who build the technology. They are the people who keynote the conferences and write the white papers and develop the frameworks and ship the work. They are, frankly, the most talented and most committed group I have ever worked with. The restructure did not break the agency. It revealed the agency. The 38 who stayed had always been the core. The other 282 had been around them, doing the work that the era we were leaving had needed done, while the core had been quietly waiting for the era we were entering to arrive.

Everyone on our team had the opportunity to stay, learn and evolve.

Thirty-eight did.

That sentence is going to be one of the most quoted things in this book. I want you to register what it means before it gets quoted. It does not mean 38 out of 320 people were good enough to keep up. It does not mean the other 282 were not. It means that 38 out of 320 people made the same bet I made on the lanai and stuck with it through eighteen months of difficult retraining, while the other 282 made a different choice for their own reasons — most of them entirely defensible reasons — and we honored that choice as fully as the operating discipline allowed.

It also means that the agency you are reading about right now — CI Web Group, the company writing this book, the partner of Daikin and Ferguson and Service Nation and over a hundred independent distributors and thousands of trades operators across the country — is the version that came out of the restructure. Not the version that went into it. The agency on the other side of the restructure is the agency. Not the one that existed before.

That is what thirty-eight did.

What It Cost

I want to close this chapter on the cost, because the cost is the part of the restructure I do not want any reader to skip.

The financial cost was significant. Severance. Client transitions. Technology investment. The opportunity cost of not bringing in PPC and SMM revenue while the new service lines were still being built. I will not bore you with the numbers because the numbers, for the purposes of this book, are not the point. The point is that I made the bet from a position of being able to absorb the financial cost, and not every operator reading this is going to be in that position when their version of this moment arrives. If you are, do it. If you are not, build the position to absorb it before the moment arrives, because the moment is going to arrive whether you are ready or not.

The operational cost was significant. The agency lost capabilities for several months that we had to rebuild. We lost institutional knowledge in some areas that we have spent the last two years recovering. We lost some client relationships that had been valuable, partly because the clients did not want to follow us into the new model and partly because the people they had been working with had moved on with them. Restructures lose more than they intend to lose. That is a cost I would budget for in advance if I were doing it again.

The personal cost was the highest of the three.

I want to tell you what that means, because it is the part of the restructure that the standard CEO-pivot narrative leaves out, and the part the reader who is about to make their own version of this move most needs to hear honestly.

I lost relationships I had spent fifteen years building. Not all of them. Most of the people I parted ways with stayed friends. But some of them did not. Some of them experienced the parting as a betrayal, and I do not blame them. The way you part with someone is something you can control. The way they choose to interpret the parting is not. I made the parts I could control as honorable as I could make them. I made my peace, privately, with the parts I could not control.

Not everyone made it.

That sentence is the four-word version of the longer story. Our entire team had to learn new systems and had to test theory and hypothesis fast. We had to rethink everything we had done. The pace was relentless. The work was unfamiliar. Some people thrived inside that environment and have built careers I am proud to have helped them build. Some people did not. Both outcomes were respectable. The agency could not move at the speed the AI era required without that sorting happening, and the sorting was not, in the end, anybody’s fault. It was the consequence of the era we are operating in.

I lost sleep. I lost weekends. I lost the ability, for a stretch of about a year, to do almost anything outside of the work. Michael was patient. The kids were patient. My mother, who was still grieving my father, was patient. The team that came with me was patient. The clients who came with us were patient. Everyone in my life was patient with me through the restructure, and I did not have the bandwidth to repay any of them in real time. I have spent the last two years repaying them. I am still repaying some of them.

I lost the version of CI Web Group that had been my home for fifteen years. The agency that came out of the restructure is, by every measure, a better agency than the one that went in. It is also a different agency. The one that went in had a particular shape, a particular cadence, a particular set of inside jokes, a particular set of relationships, a particular feel when you walked into the office. That agency is gone. The 38 who stayed have built a new one. The new one is excellent. It is also not the one that I had built up to that point. There is grief in that, and I am willing to name it as grief. The version of myself that walked into the lanai on November 30, 2022 had been the CEO of one company for sixteen years. The version that walked off the plane in Houston a week later was about to become the CEO of a different one. Both of those CEOs are me. Both of those companies are mine. The transition between them is what this chapter is about.

I would do it again.

Not because the cost was small. The cost was not small. I would do it again because the cost of not doing it would have been larger. The agency that did not restructure in 2023 is the agency that, by 2026, would have been quietly going out of business while wondering why its clients were leaving. I have watched that agency, in real time, in the trades marketing space. There are several of them. They are still operating. They are still telling their clients that AI is overhyped. They are still pumping out the same kind of work they were pumping out in 2018. They will not be operating in 2030. The cost of waiting, which I told you about in the last chapter, is the next decade of your business. The cost of the restructure was eighteen months of pain, 282 parted relationships, and a version of the company that I had to grieve. The cost of waiting would have been the company itself.

I am writing this book from the agency that came out of the restructure. The 38 who stayed are reading this draft alongside me. The relationships that survived the restructure are stronger than the relationships that preceded it. The technology stack that came out of the restructure is the technology stack that is positioning trades operators across the country for the AI era. The operating discipline that came out of the restructure is the discipline that is, three and a half years later, the most durable competitive advantage CI Web Group has.

I would do it again because the cost of not doing it would have been larger.

That is what thirty-eight did.

And that is, in operational terms, where the rest of this book starts.

There is one more thing I have to tell you before we move forward.

Everything I have just told you about the restructure happened while the rest of the trades industry was being told the opposite. The restructure was the lonely version of the right answer. The industry version was that nothing had changed, that the playbook from the last decade still worked, that AI was hype, and that contractors should stay the course.

That part of the story is the next chapter.