The B2B Event Market Is Splitting in Two

The B2B Event Market Is Splitting in Two

Part 3 of B2B Media in the Machine Age A note on timing: This post is publishing on a Friday, a break from our usual Monday/Wednesday cadence. The reason is a good one. I'll be back Monday with a piece that came directly out of a presentation I gave this week to a commercial real estate leadership group. I was invited to speak on agentic AI and its impact on commercial real estate, and it clarified something important about where agentic AI is starting to reshape the physical economy. See you Monday. A few weeks ago, Apollo Global Management announced it was acquiring Emerald Holding for approximately $1.5 billion, a 42.1% premium, and simultaneously acquiring Questex, with the intention of combining the two into a platform of roughly 160 B2B events. The deal is the largest private equity bet on B2B events in years. Apollo’s rationale was stated plainly: “As AI and digital tools rapidly expand the ways professionals connect and share information, they are simultaneously elevating the value of trusted, in-person gatherings, where industries come together to do business, build relationships, and make consequential decisions.” That is either a very smart thesis or a very expensive rationalization. I've been involved in the B2B events market for most of my career: as a CEO, as a producer and operator who has built and run hundreds of events, including major technology trade shows. What I see happening right now is not a bull market in B2B events. It is a structural bifurcation. And the M&A activity, far from being a sign that the whole category is healthy, is actually a trailing indicator of a market in the middle of a painful sorting process. Let me explain what I mean. The Bull Case Is Real, But Incomplete Start with the data that supports the optimists. The U.S. B2B trade show market exceeded $15.78 billion in 2024 and is projected to surpass $17.3 billion by 2028. The post-COVID recovery in live events has been genuine. Eighty-three percent of marketing decision-makers expect marketing investments to grow in 2026. According Content Marketing Institute research, B2B marketers rank in-person events as their most effective content distribution channel, ahead of webinars, email, and social media. Apollo is not wrong that AI is paradoxically making great live events more valuable. When an AI agent can attend a webinar, summarize a panel, and extract every lead from a virtual event in minutes, the irreplaceable asset becomes physical presence in a high-trust, high-context environment. The things that cannot be digitized: the hallway conversation, the dinner where deals actually happen, the moment when a buyer and seller read each other across a table, become more scarce and therefore more valuable. That logic is sound. The problem is that it applies to some events, not to the category as a whole. And institutional capital has a way of buying the category rather than the distinction. The pendulum swung hard toward events, and for good reason. When digital disruption hit B2B media, the path forward wasn’t obvious. Audiences fragmented. Attention scattered. Display and programmatic commoditized everything they touched. Events were the one line that didn’t behave like the rest of the business. They were tangible, defensible, and they made money in a way that felt increasingly rare. At the end of the day, events are a far simpler business. Sell a booth, sell a sponsorship, fill a room. Until they’re not. That retreat into simplicity wasn’t wrong. But it wasn’t a strategy either. It was the path of least resistance in a category that had run out of easy answers everywhere else. And a business built on the path of least resistance tends to get exposed the moment the underlying conditions shift. They’re shifting now. The Bear Case Nobody Is Talking About Look past the headlines and the data tells a more complicated story. Sixty-nine percent of B2B events leaders saw their event budgets stay flat or decrease in 2025. The share of organizers expecting budget growth in 2026 dropped to 40%, down from 70% in 2025, a dramatic deceleration in a single year. And 75% of exhibitors report pressure to reduce exhibit costs, with nearly a third feeling that pressure directly from senior leadership. Meanwhile, the supply side is severely bloated. The past five years produced a proliferation of events across virtually every B2B vertical: many of them launched during the post-COVID bounce when demand was recovering and the bar for a “viable event” was temporarily very low. The market has not yet fully repriced that excess inventory. Here is the question I would ask any P/E investor underwriting a B2B events platform right now: Of those 160 events in the combined Emerald/Questex portfolio: how many are genuine category leaders with defensible market positions and profitable unit economics, and how many are subscale shows competing for the same shrinking pool of mid-market sponsorship dollars? Scale is not a strategy. It is a starting point. And aggregating subscale assets does not automatically produce a premium platform. They produce overhead. The Venue Directory data is telling: in the UK market, total event inquiries declined 4% in 2025, while the average RFP value increased 2%. Fewer companies asking. Bigger commitments from the ones who are. That's not a market in decline. That's a market concentrating around fewer, more serious buyers. Bifurcation, not a boom. The Sorting Is Already Underway Here is what I believe is actually happening, and what I see when I look at the data alongside my own operating experience: The B2B events market is sorting itself into two distinct tiers, and the gap between them is widening. * Tier One: The Category-Defining Show. In most verticals, one or two large-scale trade shows will survive and likely strengthen. These events have established brand equity, dominant market share among exhibitors and attendees, and the network effect that makes them self-reinforcing. If your industry peers are there, you have to be there. Sponsors know it, which is why top-tier sponsorships command premium pricing that smaller shows simply cannot match. The research confirms the logic: a single Tier-1 sponsorship producing 30 ICP-fit meetings outperforms eight Tier-3 sponsorships producing a handful of meetings each. Concentration produces leverage. Spread produces noise. * Tier Two: The Subscale Middle. This is where the pain is happening. Events with a few hundred to a thousand attendees, modest sponsor rosters, undifferentiated programming, and no clear reason for a senior buyer to travel and spend two days away from the office. Many of these events were viable when the market was growing fast enough to carry marginal players. That grace period is over. Planning for large-scale events dropped 12% year over year in 2026, and the budget rationalization happening inside B2B marketing departments is accelerating the winnowing. These events do not need better marketing. They need a fundamentally different model or an exit. What is emerging in the space left by the subscale middle is something altogether different, and it may be the most interesting development in B2B events right now. The Return of the Intimate Event There is a third format quietly gaining traction that does not fit neatly into the traditional event industry model, and it deserves its own analysis. Call it the intimate conference, the executive summit, the hosted roundtable, the closed-door dinner. The format varies, but the defining characteristics are consistent: small by design (25 to 150 attendees), curated ruthlessly, content-rich, and structured specifically to create real buyer-seller engagement in an environment where both parties actually want to be in the room. These are not the scaled-down version of a trade show. They are a fundamentally different product. Sixty-three percent of event organizers report increased demand for micro-events and intimate gatherings. Planning for small, hosted events is up 59% year over year. And the research on why is unambiguous: executives who will not attend a five-hundred-person conference will often accept an invitation to a closed-door peer discussion. The signal-to-noise ratio is entirely different. One trend piece from the events industry captured the format precisely: dinners of twenty-five to thirty carefully selected peers, no stage, no slides, no formal presentations, just honest, peer-level discussion about what is actually working, guided by a respected industry leader. That format reflects a broader shift in what senior buyers actually value when they give up two days of their calendar. The content richness matters as much as the intimacy. The events that are thriving in this format are not networking events with a thin content wrapper. They are substantive: built around real intellectual engagement, genuine industry debate, and a programming philosophy that treats attendees as practitioners, not audiences. The Budget That Moved Here is the insight I have not seen written clearly anywhere, and it changes the economics of the intimate event model significantly. The conventional assumption is that B2B event sponsorship comes from the marketing budget: specifically the events or demand generation line, controlled by the CMO or VP of Marketing. That assumption is increasingly wrong for the intimate conference segment. The high-value, curated event, the executive dinner, the invitation-only summit, the closed-door roundtable, is increasingly funded from field marketing budgets. And field marketing budgets are controlled by sales leaders, not corporate marketing. This matters enormously for several reasons. First, the ROI calculus is different. A sales leader evaluating a $25,000 sponsorship of an intimate roundtable with 40 qualified senior buyers is making a fundamentally different calculation than a CMO evaluating the same investment. The sales leader wants pipeline. The CMO wants brand reach and lead volume. Intimate events are almost perfectly calibrated to the sales leader’s criteria and poorly suited to the CMO’s. When more of the sponsorship dollars for these events comes from field marketing, the events become more durable, because they are being funded by the part of the organization that measures outcomes most directly. Second, field marketing budgets are growing even as corporate marketing budgets flatten. The pressure on CMOs to demonstrate brand-level ROI from events has never been higher. The pressure on sales leaders to find efficient pipeline has never been higher. Intimate, curated events serve both, but they particularly serve sales, which is why the budget is shifting. Third, and most importantly for event operators: this means the buyer for your intimate event sponsorship is often sitting in a different seat than the buyer for your trade show sponsorship. If you are selling intimate event sponsorships the same way you sell trade show booths: to the same contacts, with the same collateral, at the same price points, you are leaving money on the table and probably losing deals you should be winning. What Agentic AI Changes About All of This The Apollo thesis, that AI makes in-person more valuable, not less, is correct in principle. But it is incomplete. The more precise version of the thesis is this: AI makes the right in-person experiences dramatically more valuable, while making the wrong ones far easier to skip. Here is why. When an AI agent can synthesize the proceedings of a three-day conference, extract every relevant insight, identify every vendor worth evaluating, and deliver a crisp briefing document, the value proposition of attending a mediocre event collapses. Why would a senior executive spend two days and several thousand dollars to attend an event whose intellectual output can be captured and distilled by an AI with a quick prompt? The answer is: they will not. And they are already rationalizing their event calendars on exactly this basis. What AI cannot replicate is presence. Physical presence in a room of thirty senior practitioners who are all wrestling with the same problems. The relationship that forms over dinner. The conversation that happens in the fifteen minutes between sessions when two people realize they should be doing business together. The trust that is built when a buyer watches a seller engage honestly with a peer-level challenge rather than defaulting to a pitch. Agentic AI does not threaten those experiences. It actually elevates their scarcity value. But it will ruthlessly expose and accelerate the decline of the events that were always surviving on convenience and inertia rather than genuine value. Takeaways For B2B event operators and producers: Your portfolio strategy needs to reflect the bifurcation, not pretend it does not exist. If you own or operate a Tier-1 show in your vertical, invest aggressively in defending and extending that position. The network effects are real and the window to cement category dominance is narrowing. If you operate subscale events in the middle tier, the question is not how to market your way out. It is whether the event has a genuine reason to exist, and if so, whether it can be repositioned as a premium intimate format rather than a smaller version of a trade show. Those are very different events and require very different operating models. The intimate format is not a budget trade show. It is a distinct product that requires different programming philosophy, different sponsor relationships, different pricing, and often a different internal champion at the sponsor company. For B2B information and media companies: Events are increasingly the one revenue line that AI cannot directly disintermediate. Your content can be summarized. Your newsletters can be replicated. Your SEO can be outranked. Your in-person community cannot be cloned. The strategic imperative is to develop a deliberate events architecture, not a collection of shows, but a portfolio designed around specific audience segments and intentional format choices. The companies that figure out how to build category-defining shows in their verticals, while simultaneously running curated intimate formats for senior practitioner audiences, are creating a defensible revenue model. The ones running a handful of subscale shows alongside declining digital revenues are in a structurally deteriorating position. For P/E investors: The Apollo/Emerald/Questex bet may prove right over time. The thesis is sound. But the execution risk is significant, and the questions worth pressure-testing are: What is the quality distribution within that portfolio of 160 events? How many are genuine Tier-1 category leaders versus subscale assets that are being carried by the platform? What is the plan for the middle tier: rationalization, repositioning, or hope? The more interesting investment thesis in B2B events right now may not be the large-scale consolidation play. It may be the premium intimate format operator who has figured out the field marketing budget unlock, built authentic practitioner communities, and is running events that senior buyers actually want to attend. That is a smaller company, a harder business to find, and a much better risk-adjusted investment. The B2B events market is not dying. But it is splitting into winners and losers faster than most of the institutional capital entering the space appears to understand. The events that survive will do so because they are genuinely irreplaceable: because the experience cannot be synthesized, compressed, or skipped. Everything else is in a slow motion decline that more M&A will not fix. The views expressed in Uphoff on Media are entirely my own. They don’t represent the opinions of any company I’ve led, any board I’ve sat on, or any investor who’s had the pleasure of debating strategy with me over the years. If something I write here sounds brilliant, I’ll take full credit. If it turns out to be wrong, I was clearly misquoted by myself. “Uphoff on Media” is published by Tony Uphoff, Founder and Managing Partner of Uphoff Advisory, LLC [https://uphoffadvisory.com/]: a strategic advisory practice for founders, CEOs, and investors in B2B media, marketing, and technology. The businesses that drive business. This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit tonyuphoff.substack.com [https://tonyuphoff.substack.com?utm_medium=podcast&utm_campaign=CTA_1]

The Vertical Intelligence Company

Part 2 of B2B Media in the Machine Age Something significant is being built from the ruins of ad-supported B2B media. Not a replacement for what was lost. Something more valuable and sustainable. Call it the vertical intelligence company. These businesses don’t look much like the B2B media companies that came before them. They’re not chasing page views and declining CPMs. They’re not dependent on programmatic advertising, platform algorithms, or the next trade show cycle. They’re building something that professional buyers, executives, and operators have always needed and rarely had: trusted, domain-specific intelligence that’s hard to find anywhere else, delivered efficiently, priced appropriately, and designed for the way business professionals actually work today. The irony is sharp. The forces that destroyed the old model are the same forces enabling the new one. The Trends That Changed the Market Three structural shifts have converged to create an opening that didn’t exist five years ago. * The professional trust deficit reached a breaking point. Business professionals have always been skeptical of general business media when it comes to decisions that actually matter. But the gap between what general media can deliver and what domain experts actually need has never been wider. When a procurement executive needs to understand how agentic AI is reshaping supplier selection, or when a payments operator needs current intelligence on cross-border settlement, Bloomberg and the Wall Street Journal are not the answer. They were never the answer. What’s changed is that the alternatives can now be built. * The cost structure for building vertical intelligence collapsed. Proprietary data collection, analysis, content production, distribution, and subscriber management used to require enterprise-scale infrastructure. That’s no longer true. A well-run team of three to five people can now build a meaningful data asset, publish at a professional cadence, reach a targeted professional audience, and manage subscriber relationships, at a fraction of the cost that would have been required a decade ago. The barriers that kept this model the exclusive domain of large research firms like Gartner and Forrester are down. * The monetization stack expanded. The old B2B media model ran on two revenue lines: advertising and events. The vertical intelligence company runs on five or six. Subscriptions. Tiered data access. Benchmarking reports. Sponsored research. Community and membership. Consulting adjacencies. None of these revenue streams is new in isolation. What’s new is the ability to stack them at small-company scale, without the overhead that used to make multi-revenue-line businesses the province of large media conglomerates. The Accelerant: Agentic AI Here is where the story gets interesting for operators and investors paying close attention. The AI moment is broadly understood as a threat to media businesses. That framing is correct for one type of media business. It’s incorrect — and actually inverted — for another. Agentic AI is devastating to commodity information. If your business model depends on aggregating, packaging, and distributing information that can be found elsewhere, AI has already begun to erode that value. Search behavior is changing. Attention is shifting. The traffic model that kept general digital media alive is under structural pressure that will not reverse. But agentic AI amplifies the value of what it cannot replicate: irreplaceable domain expertise, proprietary data, and trusted professional community. A language model can synthesize publicly available information about B2B payments trends. It cannot replace a team with deep relationships inside the payments industry, producing original research, running proprietary surveys, and hosting conversations that shape how practitioners think about the problems in front of them. The vertical intelligence company is structurally positioned to benefit from exactly the shift that is accelerating the decline of general media. That’s not a coincidence. It’s a function of what these businesses are actually built on. What This Looks Like in Practice The model is already operating. You can see it clearly in businesses that have made the transition, whether or not they would describe themselves in these terms. * PYMNTS [https://www.pymnts.com/] is not a payments news site. It’s a payments intelligence company with original data, proprietary research, and direct subscriber relationships. Its value to readers and sponsors comes from domain depth and data, not from traffic volume or display advertising. * Art of Procurement [https://artofprocurement.com/] is not a procurement podcast. It’s a professional intelligence and community platform for procurement leaders, with content, events, benchmarking, and advisory services built around a specific professional audience with real budget authority and unmet intelligence needs. * ThomasNet [https://www.thomasnet.com/?utm_term=thomasnet&utm_campaign=PB%3AG%7CNT%3ASN%7CAN%3AThomasNet%7CCN%3ABranded&utm_source=adwords&utm_medium=ppc&hsa_acc=2612745988&hsa_cam=21280982770&hsa_grp=163774556842&hsa_ad=699105707981&hsa_src=g&hsa_tgt=kwd-316144171195&hsa_kw=thomasnet&hsa_mt=e&hsa_net=adwords&hsa_ver=3&gad_source=1&gad_campaignid=21280982770&gbraid=0AAAAACzAVrtQLB2sUs-QWd7_5k1njpsLG&gclid=Cj0KCQjwi8nRBhDhARIsAHZf_pbfeZFfHVcpoWGe98wWuQYy93aqULynS79pKQUp0Ur1lURnXtnZorcaAhhCEALw_wcB], through its acquisition by Xometry, demonstrated a version of this at industrial scale: a directory and information platform that became a vertical intelligence infrastructure for manufacturing sourcing: embedding data, analytics, and transactional capability into a single professional ecosystem. These businesses share a structure. They have domain expertise that took years to accumulate. They have proprietary data that can’t be replicated by scraping public sources. They have deep, trusted relationships with a specific professional community. And they have multiple ways to monetize all three. The Implications For operators: The question is no longer how to protect ad revenue. It’s whether you have the domain expertise, audience trust, and data assets to make the transition to vertical intelligence. Many current B2B media businesses do not. The ones that do have the foundation for something more valuable than B2B media has produced in a generation. The transition requires discipline, and a willingness to reject the instinct that drove many B2B media businesses further into the trap. Faced with revenue pressure, the natural move was to go broader: more topics, more verticals, more audience segments. The logic was intuitive. More reach means more inventory means more revenue. What operators discovered was the opposite. Each step out diluted the demographic precision, the audience engagement, and the domain authority that had made them valuable in the first place. They traded their core asset for scale that turned out not to be worth much. The vertical intelligence company runs the opposite play. It goes deeper, not broader. And depth, executed with the right technology and data capability, turns out to be the mechanism for scale, not the obstacle to it. You are not trying to be everything to a broad industry. You are building deep credibility with a specific professional community and creating data and intelligence that community cannot get anywhere else. That is a different editorial strategy, a different commercial model, and a different organizational design than the one most B2B media operators are running today. For investors: The valuation framework for vertical intelligence companies should not be built on CPM multiples or traditional media metrics. The right comparables are SaaS businesses and research platforms: recurring revenue, data asset value, community lock-in, and net revenue retention. These are the metrics that reflect what these businesses actually are. The market has not fully priced this yet. Businesses that look like small B2B media companies on the surface are sometimes sitting on data assets, subscriber relationships, and domain authority that are worth considerably more than their current revenue multiples suggest. The investors who develop a sharper lens for this category before it becomes consensus will have an advantage. The Larger Irony Business professionals have always needed what the vertical intelligence company delivers. Domain-specific, trustworthy intelligence that helps them make better decisions, stay ahead of their markets, and do their jobs at a higher level. That need didn’t emerge with the internet or with AI. It has always been there. What has changed is the ability to serve it efficiently, at scale, and profitably. The technology that is collapsing one model of information business is simultaneously making it possible to build a better one. The B2B media company of the future is already here. It just doesn’t call itself a media company. The views expressed in Uphoff on Media are entirely my own. They don’t represent the opinions of any company I’ve led, any board I’ve sat on, or any investor who’s had the pleasure of debating strategy with me over the years. If something I write here sounds brilliant, I’ll take full credit. If it turns out to be wrong, I was clearly misquoted by myself. “Uphoff on Media” is published by Tony Uphoff, Founder and Managing Partner of Uphoff Advisory, LLC [https://uphoffadvisory.com/]: a strategic advisory practice for founders, CEOs, and investors in B2B media, marketing, and technology. The businesses that drive business. This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit tonyuphoff.substack.com [https://tonyuphoff.substack.com?utm_medium=podcast&utm_campaign=CTA_1]

The Relationship Was the Product

Part 1 of B2B Media in the Machine Age B2B media isn’t dying. It’s being restructured by forces most of the industry still hasn’t fully identified. Over the next three posts, I’m going to name them. This series covers how the category got here, where it’s going, and what the restructuring means for the people running these businesses right now. I’ve spent most of my career inside B2B media. Publisher, CEO, operator, investor. I watched the print market collapse and the digital transformation happen in real time. I watched companies that built durable, franchise-level businesses make decisions that looked rational in the moment and proved catastrophic over time. I was inside some of those decisions. What I’ve come to understand is that the real break wasn’t the one most people talk about. It wasn’t digital. It wasn’t the platform dependency trap. It wasn’t even the collapse of print advertising, as dramatic as that was. The break was in the relationship. For decades, the most valuable thing a great B2B media company sold wasn’t reach. It was buyer intelligence. Publishers knew their professional communities better than marketers did. That knowledge was the product. Advertisers paid for access to a publisher’s understanding of the buyer, not just access to the buyer. Digital ended that. It shifted the value proposition from intelligence to execution. From knowing the buyer to delivering impressions. It looked like progress. It was actually a trade. And most of the industry made it without understanding what they were giving up. That’s where this series starts. The Wrong Lesson The wrong lesson is the most dangerous kind. It feels like wisdom. It’s derived from real experience. It’s shared broadly enough that it becomes consensus. And it sends an entire industry in exactly the wrong direction. B2B media drew the wrong lesson from its collapse. And most of the industry is still paying for it. The revenue decline was real. The audience fragmentation was real. The structural pressure from programmatic advertising, platform dependency, and the erosion of print was real. I’ll cover the mechanics of that unwind later in this series. What I want to examine now is what operators concluded from it, why that conclusion accelerated the damage, and what the businesses that survived actually learned instead. Because the survivors learned something different. Something that looks counterintuitive on the surface but turns out to be the only strategy that makes sense once you understand what was actually lost. They learned that the product was never the content. The product was the relationship, and specifically, what that relationship was built on. The Flywheel Nobody Named For decades, the best B2B media franchises ran on a flywheel that nobody stopped to name, because it didn’t need a name. It simply worked. And it’s only in hindsight we can now see how powerful this flywheel actually was. It went like this: editorial focus and depth created genuine buyer knowledge. That buyer knowledge made B2B media companies indispensable to marketers. Not because they could execute a program that reached an audience, but because they understood that audience better than the marketer did. A publisher covering the electronics manufacturing supply chain knew procurement cycles, knew what engineers were worried about, knew which decisions were being made and by whom and when. That knowledge was the basis of the relationship. Marketers came to publishers not just to advertise, but to learn. That knowledge advantage drove advertiser investment. That investment funded deeper editorial. Which produced more buyer knowledge. The flywheel spun. The value wasn’t reach. It was intelligence. Publishers were the people who knew your buyer better than you did, and were willing to make the introduction. The relationship between B2B media and B2B marketers was, at its core, a knowledge exchange. I know something you need to know. And because I know it, you need me in the room when you’re making decisions about how to reach this market. That’s not a vendor relationship. That’s a strategic partnership. And it’s the thing that justified premium pricing, long-term contracts, and the kind of institutional loyalty that made B2B media franchises durable across decades. What Digital Actually Changed Here’s what’s usually said about what digital did to B2B media: it disrupted the print revenue model, it commoditized reach, it introduced programmatic pricing, and it handed audience aggregation power to platforms. All of that is true. But it’s not the root cause. The root cause is that digital marketing fundamentally changed the nature of the value relationship. It shifted the marketer’s core question from what do you know about my buyer? To what can you execute for me? And once that question changed, everything downstream changed with it. In the pre-digital model, the marketer needed the publisher’s intelligence to make good decisions. In the digital model, the marketer had dashboards, attribution tools, programmatic platforms, and intent data feeds. The publisher’s buyer knowledge: built painstakingly through years of editorial investment, suddenly looked less like a strategic resource and more like one input among many. Or worse, like something the marketer could approximate themselves. The relationship didn’t erode. It was reclassified. Publishers went from strategic partners who knew the market to execution vendors who could deliver eyeballs. And execution is a commodity business. You can always find someone who will run the program cheaper. Digital didn’t just change the economics of B2B media. It changed what B2B media was being paid for. And that’s a much harder problem to solve than a CPM collapse. The content syndication era made this concrete. The thesis behind pay-per-lead programs — TechTarget built an entire company on it — was that documented purchase intent could substitute for the buyer knowledge publishers had historically provided editorially. If a contact downloaded a white paper, you had evidence of interest. That was something. But it commoditized fast. Because intent signals don’t convey buyer intelligence, they generate lists. And lists can be bought from a dozen vendors. The publisher was no longer in the room helping the marketer understand the buyer. They were handing over a spreadsheet and moving on. Even the smarter digital adaptations never fully restored what had been lost. The flywheel didn’t slow down. It stopped cold. The Wrong Lesson, Compounded When revenue started declining, the instinct across most B2B media businesses was to grow the addressable market. More topics. More verticals. More content formats. More audience segments. If the core audience wasn’t generating enough revenue, find more audiences. The logic tracked on a spreadsheet. Reach drives inventory. Inventory drives advertising revenue. More audience equals more revenue. What operators discovered, slowly, then all at once, was that each step out diluted the thing that had made them valuable in the first place. Demographic precision eroded. Engagement declined. The domain authority that had justified premium pricing started to look like every other trade publication. The buyer knowledge that had been built over years in a specific vertical couldn’t survive dilution into adjacent markets. Advertisers noticed before publishers did. The businesses that went broadest to survive ended up the most exposed. They had traded their one defensible asset: deep knowledge of a specific professional community, for volume that turned out to be worth nothing at scale. This is the lesson the collapse was supposed to teach. Most of the industry drew the opposite conclusion. What the Survivors Actually Learned The businesses that came through the collapse, not just surviving but building something sustainable, did something that looked irrational from the outside. They went deeper instead of broader. They doubled down on domain expertise at the exact moment the market seemed to be rewarding scale. They learned three things that the broader industry missed. Focus is the mechanism for scale, not the enemy of it. When you go deep enough in a specific vertical, the audience you build is more valuable per member than any broad audience can be. A readership of 50,000 senior procurement professionals is worth more to the right advertiser than 5 million general business readers. The math of depth beats the math of breadth once you stop measuring impressions and start measuring influence. Technology doesn’t change the value of buyer knowledge, it changes how you build and deliver it. The vertical intelligence model isn’t a nostalgia play for the print era. It’s a recognition that the original value proposition: we know your buyer better than you do, is more achievable now than it ever was. First-party data, behavioral signals, proprietary research, AI-assisted synthesis: the tools for building genuine buyer intelligence have never been more powerful. The question is whether publishers will use them to rebuild the relationship or just generate more content. The relationship has to be the product again. This is the hardest shift. It requires rebuilding sales motions, repricing the offering, and reconstituting the trust that was traded away during the programmatic era. But it’s the only path to a business that isn’t perpetually competing on execution cost. The vertical intelligence model isn’t a new idea. It’s the original idea, rebuilt for the current era, with data, AI, and workflow integration as the delivery mechanism instead of editorial alone. The Open Question for Operators The vertical intelligence model is not available to every B2B media business. It requires a foundation that not every operator has: genuine domain expertise, a professional audience with real workflow needs, and the organizational willingness to sell intelligence access rather than impression inventory. The honest diagnostic question for any B2B media operator right now is not how do we grow our audience? It is: do we still know something specific about a specific professional community that that community cannot easily find elsewhere? If the answer is yes, the path to rebuilding the strategic relationship is more accessible than it has ever been. If the answer is no, if the business traded its domain authority for breadth that didn’t pay, the recovery is harder. But the diagnosis at least is clear. The flywheel that built the best B2B media franchises was always powered by knowledge. Digital stopped it by changing what knowledge was worth. The question now is whether the industry will rebuild it with better tools, or spend another decade competing on execution in a race it was never going to win. The category isn’t just being disrupted. It’s being reinvented. In Part 2, I’ll lay out what the leading B2B media companies are actually becoming, and why the ones that get there first will be very difficult to displace. The views expressed in Uphoff on Media are entirely my own. They don’t represent the opinions of any company I’ve led, any board I’ve sat on, or any investor who’s had the pleasure of debating strategy with me over the years. If something I write here sounds brilliant, I’ll take full credit. If it turns out to be wrong, I was clearly misquoted by myself. “Uphoff on Media” is published by Tony Uphoff, Founder and Managing Partner of Uphoff Advisory, LLC [https://uphoffadvisory.com]: a strategic advisory practice for founders, CEOs, and investors in B2B media, marketing, and technology. The businesses that drive business. This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit tonyuphoff.substack.com [https://tonyuphoff.substack.com?utm_medium=podcast&utm_campaign=CTA_1]

When AI Builds Itself: What the Anthropic White Paper Means for B2B Leaders

The most useful thing I can offer on the AI disruption isn’t a prediction. It’s a pattern. One I’ve lived through before, at scale, in this sector. I was in the front seat of the roller coaster. At the time, I was running a fast-growing, $300 million print-centric B2B media business, and by every metric, we were succeeding. Then the Internet arrived. Not gradually, or at least it didn’t feel that way from the operator’s seat. One quarter you’re managing circulation strategies, analyzing market share reports, and launching events. The next, employees are stopping you in the hallway, the board is asking pointed questions, and customers want to know what you’re doing “about the Internet.” I had covered technology disruptions for years. The brands I was running served technology professionals. I understood disruption conceptually. What I had never done was lead a business through one in real time, with real consequences, with people’s livelihoods and a board’s confidence depending on decisions I had to make before I had the answers. No HBR article prepared me for it. No academic framework made it manageable. What got me through it — and what I learned at real cost — was hard, experiential learning. The ability to stay calm when the pressure was extraordinary. The ability to move forward before the path was clear. The ability to hold the organization together while simultaneously rebuilding what the organization did, what it sold, and what it was worth. That moment taught me more about leadership than anything before or since. Let me be clear about something before we go further. If you’re reading this expecting another dramatic proclamation about the impact of artificial intelligence, stop here and go read something else. There is no shortage of that content. AI is saturating every media channel, every business conference and every board agenda. The volume of coverage is inversely proportional to the operational clarity it produces. Most of it is written by people who are analyzing AI from the outside and with minimal practical insight. Very little of it is written by people who have actually led businesses through a technology disruption of comparable scale. I have. And that is the lens for everything that follows. For readers who are digital natives and have built their entire careers in a world where the Internet already existed — which is at least half of you, and probably more — the Internet disruption is history, not lived experience. Some of you were in the early stages of your careers when it happened, not yet in leadership positions where the decisions landed on your desk. I’m not telling you this to establish context. I’m telling you this because the playbook from that disruption is the closest thing to a field guide for this one. The Internet didn’t just change how B2B media companies distributed content. It restructured what content was worth, who would pay for it, and what a media business fundamentally was. The companies that survived and grew were the ones that recognized they were in a different business than the one they thought they were in, and acted on that recognition before circumstances forced it. The companies that didn’t survive treated the Internet as a distribution channel for what they were already doing. They optimized the wrong thing until it was too late to optimize the right thing. We are in that moment again. And it may be more disruptive still. Which brings us to the Anthropic white paper, and why it matters more than the coverage it has received. The Anthropic Institute titled it "When AI Builds Itself." [https://www.anthropic.com/institute/recursive-self-improvement] That title is the thesis. This is not a technology document, and it is not another breathless dispatch from the AI hype cycle. It is a document that describes, in hard data, the operating environment you are already living in, whether your organization has named it yet or not. If you lead a B2B media, marketing, or technology company — or invest in one — the questions it raises are not theoretical. They are arriving in your inbox, your board meetings, and your leadership team meetings right now. It is not AI theater dressed in business language. It is a field guide, written from the operator’s seat, for the decisions in front of you. What the Paper Actually Says: Three Things That Matter Most AI commentary asks you to imagine the future. This paper shows you the present. Anthropic is documenting, in hard operational data: code output per engineer, task completion rates, productivity multiples measured against human performance, what AI is already doing inside one of the most advanced technology organizations in the world. Three findings have direct consequence for every leader in this sector. First, the development productivity compression is not incremental. It is structural. As of May 2026, more than 80% of the code merged into Anthropic’s own codebase was authored by Claude. Before Claude Code launched in early 2025, that number was in the low single digits. The result: in the second quarter of 2026, the typical Anthropic engineer was merging 8x as much code per day as they were in 2024. In a March 2026 internal poll of 130 Anthropic employees, the median respondent estimated they were producing around 4x as much output with AI assistance as they would have produced without it. Read that again. 8x more code per engineer per day. 4x overall output. At the most sophisticated AI development organization in the world, operating at the frontier of what these systems can do. For B2B technology-enabled businesses: companies that use technology to create, develop, and distribute their products and services but are not selling technology directly, this is the most consequential finding in the paper. The engineering backlog that every one of these businesses has been managing as a permanent condition of operating is not a permanent condition. It is a constraint that is beginning to lift. And when that constraint lifts, the question is not how to go faster on the existing roadmap. It is what the roadmap looks like when the binding constraint of the past decade no longer applies. Second, the capability curve is accelerating faster than most executives have internalized. The paper documents that the length of tasks AI can reliably complete on its own has been doubling roughly every four months. In March 2024, Claude could complete software tasks that take a human roughly four minutes. By 2026, that horizon had extended to 12-hour tasks. The paper projects that tasks taking a skilled person multiple days could come into range before the end of this year. Tasks taking weeks could follow in 2027. This is not a projection from an optimistic analyst. It is a measured trend line from observed performance data, documented by the organization building these systems. The pace of change embedded in these numbers is faster than most strategic planning cycles. Which means most organizations are planning against a capability baseline that is already out of date. Third, the paper names three possible futures: and the one most relevant to B2B operators is already underway. Anthropic outlines three scenarios for what comes next: the trend stalls at current capability levels; AI labs see compounding efficiency gains while humans retain direction-setting; or AI achieves full recursive self-improvement. The paper is direct that scenario two — compounding efficiency gains with human oversight — is where the evidence points today. And the operational implication of scenario two is named explicitly: a 100-person company can increasingly do the work of a 1,000-person one, because each employee sits atop a pyramid of AI agents. That sentence is not a marketing claim. It is a structural shift in competitive dynamics. And it is happening now. The All-Too-Human Reality Before the frameworks. Before the action items. First, this. The decisions described in this post are not technology decisions. They are career decisions. Organizational decisions. In some cases, identity decisions, for leaders who have built their professional lives around capabilities that are being restructured around them. The wave of senior leadership departures we are already seeing: Adobe, Walmart, and others whose CEOs have cited AI transformation as a reason for stepping aside, is being reported as a business story. It is also a deeply human story. Experienced, successful leaders concluding that they are not the right person to finish what they started. That takes a form of self-awareness and courage that is genuinely rare, and genuinely hard. I know what it feels like to run a business through a technology disruption where the ground is shifting and the answers are not yet available. The pressure is real. The isolation is real. The temptation to project more confidence than you feel, to buy time with reassuring language while privately figuring it out, is entirely understandable. I have been there. What I learned is that the leaders who navigate disruption well are not the ones who pretended it was simple. They are the ones who stayed honest about the uncertainty while remaining decisive about the next step. Those are compatible. They just take practice. With that said: here is the field guide. For CEOs: The Two Decisions That Cannot Wait Decision One: Are you the right person to finish this? This is the question nobody is asking directly in your organization, which means you have to ask it yourself. It is not a referendum on your career or your competence. The leaders stepping aside at Adobe, Walmart, and elsewhere are not failures. Several of them are arguably demonstrating their greatest leadership act: the clarity to recognize that a different phase of a company’s life requires a different profile of leader, and the integrity to act on that recognition rather than hold the seat. The honest version of this question has three parts. Ask yourself each one directly. Do you have genuine fluency with AI? Not enthusiasm for it. Not familiarity with the vocabulary. Actual working knowledge of how these systems operate and where they create and destroy value. Do you have the appetite for genuine business model transformation? Not the efficiency overlay that is insufficient. The fundamental rethinking of what your business is and what it sells. Do you have the runway to see it through? Personal energy. Board confidence. Team trust. This is an 18-to-36-month transformation, not a 6-month initiative. The Anthropic paper puts this in sharp relief. The organizations that are capturing the gains described in this paper are not doing so by running AI pilots. They are rebuilding how work gets done at a fundamental level. That requires a leader who can hold the organization through sustained uncertainty. Not someone who manages to a quarterly plan and waits for the path to clarify. If the answer to all three questions is yes, your job is to lead with full commitment and full transparency. If the answer to any of the three gives you pause, that pause is worth thinking about before circumstances force the conversation. Decision Two: How do you set expectations that are simultaneously honest and stabilizing? You have four distinct constituencies and they need four different conversations. Conflating them is where leaders make the most consequential mistakes. * Employees need to hear the honest version before they read the trade media version. Your people are not naïve. They can see what is happening around them in the sector. The Anthropic paper documents that AI is already handling more than 80% of code production at one of the world’s leading technology organizations. People doing knowledge work; writing, analysis, research, production, are drawing their own conclusions about what that trajectory means for their roles. Vague reassurance does not calm that anxiety. It amplifies it, because it signals that leadership either doesn’t know or won’t say. What employees need is an accurate picture of what is changing, a genuine account of what it means for their roles, and a believable plan for how the organization is going to navigate it. That plan does not need to be complete. It needs to be real. * Boards need a different frame entirely. Most boards are running on mental models formed in the last disruption cycle. They are simultaneously over-excited about AI as a narrative and under-equipped to evaluate it as a business reality. The paper’s finding that Anthropic engineers are merging 8x more code per day than two years ago is exactly the kind of concrete, operational data point your board needs. Not as a benchmark to match, but as evidence of the pace of change they are governing through. Your job with the board is not to match their enthusiasm or manage their anxiety. It is to be the most credible voice in the room about what AI actually means for your competitive position, your cost structure, and your product roadmap over the next 18 months. Specificity is your currency. Vague AI optimism is not. * Investors, particularly in PE-backed companies with defined fund timelines, need the transformation timeline mapped against the investment thesis clearly and honestly. Efficiency gains from AI adoption can show up in 6 to 12 months. Revenue model transformation: new intelligence products, subscription conversion, data licensing, is an 18-to-36-month story. If those timelines do not align with fund structure, that is a conversation to have now rather than at the next quarterly review. * Customers want proof, not vision. The AI positioning saturating the market right now ranges from genuinely substantive to transparently hollow, and your customers can usually tell the difference. What they need from you is specific capability milestones, honest timelines, and a clear account of how your AI investments translate into value for them. The vendors who earn trust in this period will be the ones who under-promised and over-delivered. For Operators: Three Decisions About How the Work Gets Done Decision One: The development bottleneck is lifting. Are you positioned to act? The Anthropic paper documents something that should fundamentally change how every B2B technology-enabled business thinks about its product development process. When AI systems can write production-quality code at 8x the previous rate, and when that code is passing the review standards of the world’s best engineers, the operating assumption that has governed every technology roadmap for the past decade — that engineering capacity is the binding constraint — is no longer reliable. This is not a small adjustment. For most B2B technology-enabled businesses, the operating rhythm has been: more demand than we can build for, so we prioritize ruthlessly and leave significant product opportunity on the table. The AI-driven productivity shift does not eliminate prioritization. It changes the calculus for what is worth building at all. Workflow improvements that were previously uneconomical become viable. Vertical-specific modules for individual enterprise customers become feasible. The long-deferred infrastructure work that has been on every product roadmap for three years finally gets addressed. The paper notes that Anthropic engineers are now using AI to fix problems that had accumulated for years, including one case where Claude shipped over 800 fixes that reduced a class of API errors by a factor of one thousand, work an engineer estimated would have taken a human four years to complete. The question for operators is not “how do we use AI to go faster on our existing roadmap.” It is “what does the roadmap look like when the constraint we have been managing for ten years is no longer the binding constraint.” Those are different questions and they produce different answers. Decision Two: Is your implementation model a competitive advantage or a liability? The Anthropic paper describes a world where AI agents can work autonomously for hours on complex, open-ended tasks: tasks with no clear specification, where the approach itself has to be figured out along the way. Claude’s success rate on the most open-ended tasks reached 76% in May 2026, up 50 percentage points in six months. For B2B technology vendors, this has a direct implication for professional services revenue. If AI agents can execute configuration, data migration, and workflow setup tasks that currently require billable professional services hours, then implementation revenue, a significant line item in many B2B technology contracts, compresses or disappears. Some operators will experience this as margin erosion and fight it. The sharper operators will get ahead of it by treating faster time-to-value as the selling point it actually is. Shorter implementation cycles mean shorter time to demonstrated ROI, which shortens sales cycles, improves renewal rates, and reduces customer acquisition cost. The math on giving up implementation revenue in exchange for those outcomes is often compelling. The operators who do this analysis now rather than after the revenue line is already under pressure will be in a better position. Decision Three: What not to do, and what to do now The mistakes being made at scale right now in this sector are worth naming plainly, because they are tempting mistakes made by smart people under real pressure. * Premature restructuring announced before the capability exists. Declaring an AI transformation in a board presentation or a customer communication without the underlying product reality to support it. This is not a positioning mistake. It is a trust mistake, and it is very difficult to recover from once customers and employees have seen the gap between the announcement and the reality. * Cutting the talent you will need in 18 months. The current wave of AI-driven workforce decisions contains real efficiency gains and genuine AI-washing in roughly equal measure. The leaders who will regret decisions made in this period are the ones who eliminated domain expertise in the name of efficiency. Editorial judgment. Deep customer relationships. Vertical market knowledge. These are exactly the inputs that make AI-generated intelligence valuable rather than generic. Cut them now and you will spend the next three years trying to rebuild what you gave away. The Anthropic paper is instructive on this point. The human role that remains irreplaceable, even at the most AI-intensive organization in the world, is research taste and judgment. Choosing which problems matter. Knowing which results to trust. Recognizing when an approach is a dead end. That same judgment, applied to your vertical market, is your most defensible asset. Protect it accordingly. * Ignoring the mid-level leadership layer. CEOs get the attention. Investors get the attention. The VP of Product, the Chief Revenue Officer, the Managing Editor, these are the people who will execute the transformation or block it. They are navigating this period with almost no guidance designed for them. The operators who invest in their mid-level leadership now, who bring them into the strategic conversation rather than cascading decisions down to them, will have a significant execution advantage. What to do now: Audit your product roadmap through the lens of a changed development constraint. Identify your implementation revenue exposure and model the unit economics of faster time-to-value as an alternative. Inventory your proprietary data assets, genuinely, not aspirationally. And protect the domain expertise in your organization as the strategic asset it is, because it is what will make your AI-generated intelligence valuable rather than generic. For Investors: The Filter That Matters Now The Three-Characteristic Test The PE-owned B2B media and marketing landscape contains a wide distribution of asset quality, and AI is going to widen that distribution significantly. The Anthropic paper’s scenario two: compounding efficiency gains, with a 100-person company doing the work of a 1,000-person one, is not a uniform benefit across the sector. It is a capability multiplier. And what it multiplies matters enormously. Some companies in this sector are being structurally disintermediated. Media brands whose primary value was information access. Marketing services businesses whose primary value was human execution of repeatable tasks. Research businesses whose primary value was data aggregation rather than actionable intelligence. For these companies, AI does not create a rescue opportunity. It accelerates the pressure. The companies that can use this moment to break out share three characteristics. One: A defensible proprietary data asset. Not licensed data. Not aggregated public data. Behavioral data, transaction data, intent data that comes from a genuine and exclusive relationship with an audience or customer base. Two: A trusted audience or customer relationship. The kind of relationship where the brand is the reason people show up, not just the container for the information. Three: A leadership team willing to pursue genuine business model transformation. Not an efficiency overlay on a declining model. A fundamental rethinking of what the business is and what it sells. The efficiency overlay is the more common response because it is easier to underwrite, easier to model, and easier to defend in a board meeting. It is also insufficient. The B2B media and information companies that navigate this period successfully will be the ones that use AI not to do the same thing cheaper, but to do something fundamentally more valuable: moving from content delivery to intelligence execution, from audience access to workflow integration, from quarterly reports to continuously updated decision support. The Timeline Reality The Anthropic paper’s capability trajectory has a direct implication for investment timelines that most sponsors are not explicitly addressing. If AI systems can complete tasks that take humans days by the end of 2025, and tasks that take humans weeks in 2027, then the window for business model transformation is not a three-to-five year horizon. It is compressed. Companies that begin genuine transformation now have a meaningful advantage over those that begin in 18 months. Not because the technology will be unavailable, but because the organizational learning curve, the product development cycle, and the audience trust that makes intelligence products valuable all take time to build regardless of how capable the underlying AI systems become. The timeline reality for PE sponsors is specific and needs to be addressed. Efficiency gains show up in 6 to 12 months. Revenue model transformation is an 18-to-36-month story. Most fund structures were not built around that gap. The result is pressure to declare transformation victories on an efficiency story that acquirers will correctly identify as incomplete. That misalignment is a risk factor in its own right. Sponsors and portfolio company leadership need to have an explicit conversation about whether the fund structure can actually support genuine transformation. Most are not having it. What AI-Native Leadership Actually Looks Like in Diligence The standard diligence question about AI readiness: “how are you using AI in your business”, produces answers that are almost universally unreliable, because every management team has a version of the answer that sounds good and very few have the substance behind it. Better questions: What proprietary data does this business own that gets more valuable as AI systems get better at synthesizing it? What is the gap between the intelligence this business could produce from its data and the intelligence it is currently producing? Has the leadership team demonstrated the ability to move through ambiguity and make structural decisions rather than incremental ones? And specifically: who on this team has led through a prior technology disruption cycle at the operator level, not the advisory level? That last question matters. The Anthropic paper is clear that the human role that persists, even as AI takes over the execution of complex tasks, is judgment about which problems are worth solving. That judgment, in a business context, comes from having made hard calls under real pressure, with real consequences. It cannot be learned from a course or acquired by reading white papers. The difference between leaders who have run businesses through disruption and leaders who have studied businesses through disruption is significant, and it is not always visible in a management presentation. The Vertical Business Intelligence Window For small and mid-sized B2B media and information companies, the Anthropic paper describes something specific and time-bounded: the infrastructure cost of building a sophisticated intelligence product has collapsed. The paper documents that Anthropic engineers are now using AI to complete work that would have taken humans years. Not through artificial intelligence as a concept, but through AI agents executing complex, multi-step workflows autonomously over hours or days. The same capability that is transforming software development at Anthropic is available to a focused B2B media company building a vertical intelligence product today. Five years ago, building a genuine intelligence product on top of a vertical data asset required a data science team, an engineering team, and capital that most B2B media companies did not have. Today, a focused team of two or three people can build and deploy an agentic pipeline that ingests, synthesizes, and surfaces vertical market intelligence at a level of sophistication that would have required an enterprise analytics operation in 2020. The capability documented in the Anthropic paper is not confined to Anthropic. It is available now, at a cost structure that has never existed before. This matters because the moat in B2B media has always been the proprietary relationship with the audience: the behavioral signals, the transaction data, the intent signals that come from a community that trusts the brand. What was missing was the productization capability. Agentic AI closes that gap. The companies that recognize this and move are positioned to make the long-predicted but never-quite-arrived shift from advertising revenue to audience intelligence revenue. The companies that treat AI as a content production efficiency tool are going to miss the more consequential opportunity. The window is open. The Anthropic paper makes clear it will not remain open indefinitely. The large players are moving, the capability is diffusing, and the organizations that build institutional AI experience now are accumulating a compounding advantage over those that wait. The advantage small and focused B2B media companies have right now is speed, domain specificity, and audience trust. Those advantages are real and durable, but they require action to be realized. What to Do Now If you are a CEO: Have the honest internal conversation before the board forces it. Assess your own AI fluency with the same rigor you would apply to assessing any other capability gap in your organization. Begin the four-constituency communication sequence now, starting with your employees. They need to hear the honest version from you before they read the trade media version. And make the organizational decision about AI leadership explicit: who in your organization owns this transformation, with real authority and real accountability, not as an addition to an existing role but as a primary responsibility. If you are an operator: Audit your product roadmap through the lens of a changed development constraint. The 8x productivity figure from the Anthropic paper is a leading indicator of where this goes, not a ceiling. Identify your implementation revenue exposure and model the unit economics of faster time-to-value as a strategic alternative. Inventory your proprietary data assets, genuinely, not aspirationally, and build toward the intelligence product your vertical market will pay for. Protect the domain expertise in your organization as the strategic asset it is. If you are an investor: Apply the three-characteristic filter to every portfolio company: proprietary data asset, trusted audience relationship, leadership capable of genuine transformation. Have the explicit conversation with each portfolio company about transformation timeline versus fund timeline, and be honest about the misalignment where it exists. Build an AI-native leadership evaluation framework before your next diligence process, because the standard questions are producing unreliable answers. The capability trajectory in the Anthropic paper compresses your planning horizon. Act accordingly. The Path Forward I came out the other side of the Internet disruption a better leader, a sharper operator, and far more clear-eyed about what the job actually requires. What I know now is that the leaders who navigate disruption well share a common characteristic. They do not wait for clarity before they act. They move toward the uncertainty rather than away from it, they stay honest with the people depending on them, and they build the organizational capacity to learn and correct in real time rather than the organizational theater of having all the answers. The Anthropic white paper is telling you, in hard data, that the ground is shifting. The question is not whether this disruption is real. The paper answers that. The question is what kind of leader you are going to be in response to it. That is the most important decision of all. And it is entirely yours to make. The Anthropic Institute white paper “When AI Builds Itself” is available at anthropic.com/institute/recursive-self-improvement. The views expressed in Uphoff on Media are entirely my own. They don’t represent the opinions of any company I’ve led, any board I’ve sat on, or any investor who’s had the pleasure of debating strategy with me over the years. If something I write here sounds brilliant, I’ll take full credit. If it turns out to be wrong, I was clearly misquoted by myself. This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit tonyuphoff.substack.com [https://tonyuphoff.substack.com?utm_medium=podcast&utm_campaign=CTA_1]

How Do We Move?

Last week’s post, “The AI Layoff Myth" [https://tonyuphoff.substack.com/p/the-ai-layoff-myth] generated a ton of discussion: on the blog, on LinkedIn, and in private conversations with operators and executives across the industry. That response pointed directly to the question I'm being asked most in my advisory practice right now. This post is my answer, and the framework I've built around it. “How do we move?” That question comes up in nearly every conversation I have with founders, CEOs, executives and investors in B2B media, marketing and technology right now. Agentic AI is the context. The answers cluster into two camps. The first camp has committed. They are mandating adoption, setting deadlines, measuring usage. The second camp is interested but cautious. They are watching, waiting, gathering information before making a move. Both camps are making the same mistake. The Phase Problem In 1962, a sociologist named Everett Rogers published “Diffusion of Innovations”, a landmark study of how technologies actually move through organizations. Rogers wasn’t interested in which companies adopted first. He was interested in why adoptions fail. His finding: they fail because leaders misidentify which phase they’re in and manage accordingly. Rogers mapped five phases every meaningful technology moves through inside an organization: knowledge, persuasion, decision, implementation and confirmation. Each phase requires different leadership behavior. Skip one, and the implementation collapses, regardless of how good the technology is or how much budget is behind it. The mandate camp is running implementation plays before their organizations have completed persuasion. Compliance isn’t adoption. People follow mandates and wait them out. The wait-and-see camp has convinced themselves that accumulating knowledge is the same as strategy. It isn’t. Knowledge without structured experimentation is just informed inaction. Additive Versus Substitutive Every prior technology wave B2B executives have navigated was additive. Cloud made existing infrastructure cheaper. SaaS made existing workflows faster. Programmatic made existing media buys more efficient. Mobile extended existing experiences to a new screen. Additive technologies reward the fast follower. You can watch others absorb the implementation costs, learn from their mistakes and move when the path is clearer. The workflows you’re augmenting will still be there when you arrive. Agentic AI is substitutive. It doesn’t augment workflows. It replaces them. The research workflow. The content production workflow. The lead qualification workflow. The procurement analysis workflow. These are not getting faster or cheaper. They are being rebuilt from different starting assumptions. In a substitutive cycle, the fast follower strategy breaks down. By the time the wait-and-see executive is ready to move, the workflow they intended to augment no longer exists in its prior form. The organizational muscle memory built around that workflow is a liability, not an asset. This is what’s different. Not the technology itself, executives have managed technology transitions before. The substitutive nature of the shift is what changes the strategic calculus. Trialability: The Argument Against Waiting Rogers identified trialability as one of the strongest predictors of adoption velocity. The easier it is to experiment with a technology on a limited basis, before full commitment, the faster it moves through an organization. Agentic AI has near-zero trial cost right now. That is the fact the wait-and-see camp is underweighting. You do not need a platform decision, a vendor contract or a change management initiative to begin. Here is where to start: * Pick one workflow. Not a strategy. One specific, bounded workflow that your team runs repeatedly: a weekly report, a competitive brief, a content outline, a prospect research summary. Scope matters. Broad experiments produce ambiguous results. * Run it in parallel. Have your team complete the workflow the existing way and with an agentic tool simultaneously for 30 days. You are not replacing anything yet. You are generating comparable data. * Measure what actually changes. Not sentiment. Time to completion, output quality as evaluated by the person who normally does the work, and error rate. These three metrics will tell you whether substitution is real in your specific context. * Expand from evidence. One workflow becomes two. Two becomes a workflow audit. A workflow audit becomes a genuine adoption roadmap: phase-aware, sequenced, grounded in your organization’s actual experience rather than vendor promises or competitor anxiety. This is not a technology project. It is an operational experiment. The cost of running it is minimal. The cost of not running it compounds. The Real Question How do we move? The answer isn’t mandate and it isn’t wait. Rogers spent a career documenting what actually works: structured experimentation, phase-aware leadership and honest measurement. The organizations that navigate this transition well won’t be the ones that moved fastest or most cautiously. They’ll be the ones that moved deliberately, and started sooner than felt comfortable. The workflow you’re waiting to augment is already being rebuilt somewhere else. The views expressed in Uphoff on Media are entirely my own. They don’t represent the opinions of any company I’ve led, any board I’ve sat on, or any investor who’s had the pleasure of debating strategy with me over the years. If something I write here sounds brilliant, I’ll take full credit. If it turns out to be wrong, I was clearly misquoted by myself. This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit tonyuphoff.substack.com [https://tonyuphoff.substack.com?utm_medium=podcast&utm_campaign=CTA_1]