コンテキストサプライチェーンにおける位置

Early in my career, I worked at a company where the CEO would host a monthly AMA for all the employees. One time I’ll never forget, someone asked him a softball question that was essentially “What career advice do you have for us?”The CEO totally flubbed it.Essentially, he said if you want to be really successful you should be either building the product or selling the product.Oof. There were hundreds of people listening who weren’t in Product or Sales. People in Finance, HR, Ops and Services who thought they were part of the mission, thought they were equal in the eyes of their leader, and instantaneously realized the CEO didn’t value their work. He viewed them as a cost to manage.I’ve thought about that moment a lot over the years. While it wasn’t great leadership, I think it was good advice. It’s not even controversial. It’s Business 101-level stuff. CEOs just don’t usually say it out loud to all their employees.Peter Drucker, widely considered “the man who invented management”, said it succinctly in the 1950s:“Because the purpose of business is to create a customer, the business enterprise has two--and only two--basic functions: marketing and innovation. Marketing and innovation produce results; all the rest are costs.” It’s not that the people in other roles aren’t incredibly talented, or critical for the business to function. It’s just that they’re…costs.The vast majority of these roles are what I would call a “coordination tax” the business needs to pay in order to bring their product to market.Now, because of AI, many companies are asking themselves the obvious question: Why are we still paying the coordination tax?Every organization runs on what I like to call a “context chain”. At the front of the chain are the people who create the original context: the actual product, the service model, the media, the strategy etc. All the way on the other side are the people who put the product into the hands of customers. And in between these two groups are an enormous number of people who carry the product, service, or decision context from one point to the next, repackaging it along the way for the next person in the chain. Every job in the context chain can be grouped into one of three categories: 1. Innovators. The people generating the original context. Engineers writing the code, researchers running the experiments, leaders making the strategic bets. They produce the “thing” that flows down the context chain and ultimately to the customer. 2. Context Carriers. Carriers take context from one part of the business to another. Their job is to contextualize the product for a different audience along the way. They summarize, reformat, enable, document, coordinate, QA, and train different people in the business on the new product. All of these people represent the coordination tax. 3. Marketers. The people who work to get the product into the hands of actual customers. The folks who work in sales, marketing, and partnerships. Note: There’s a fourth bucket of compliance, legal, and finance people who keep the business from blowing up. Finance people I know you do more and are strategic, don’t yell at me.If you’ve never worked in a large organization, it’s hard to articulate how much of the work being done is just carrying context from one team to another.Let’s use launching a new software product as an example. The product team builds the product. That’s innovation. Now here is the coordination tax the business pays to deliver the product to their customers:Someone needs to document how the product works, where it breaks, and what the known issues are.The support team needs to learn how to troubleshoot it, which means the documentation gets repackaged into support playbooks.The implementation team needs to learn how to set it up. They need implementation guides and training materials.Account management needs to learn how to drive adoption and upsell, so that’s talk tracks and value frameworks.Sales needs to learn how to sell it. They need pitch decks, competitive battlecards, demo scripts.Marketing needs to position it in the marketplace and run campaigns, which means messaging docs, landing pages, launch plans.The partner network needs to learn it too, so everything gets repackaged one more time for an external audience with completely different incentives.That’s seven rounds of recontextualizing the same fundamental information for different audiences. Hundreds of people are employed in context-carrying roles at a mid-size company. At a large business it can be thousands and thousands of people in the context chain.This cascade isn't just slow and expensive, it doesn't even work well. The product team spent months building a specific product using a massive amount of research with deep understanding of the customer problem and market opportunity behind every decision. All of that context lives in the heads of the people who built it, and now it needs to pass through seven or eight handoffs. But if you’ve ever worked at a software company, you know that the product people absolutely. hate. documentation. You can yell at them. The CEO can yell at them. They. will. not. do. the. documentation. So many companies have whole teams to gather this context into the “big package” which has many names at many companies but is basically “here is all the documentation we could find about this launch.” So even the “big package” is incomplete, and it’s messy because a human did it and we aren’t perfect at synthesizing information with very large context windows (spoiler alert: do you know what is absolutely incredible at synthesizing information using a very large context window?).Anyways, the PMM who writes the positioning document understood most of what the product team intended, but not all of it. The enablement lead who builds the training has their own spin on what’s critical to understand. The sales manager who trains their team added their own context. By the time the rep is on the phone with the customer, they’re four or five handoffs removed from the original intent, working from a version of the story that’s morphed into a pitch quite different than the original innovation intended. The point is that this is a structural problem. It is embedded into the operating system of the business. Until AI came along…If you’ve used AI even a little, you know that AI is exceptional at synthesizing information from disparate sources and formats and then contextualizing it to the appropriate end user.With AI, the information doesn’t degrade with each transfer through the human context chain. Instead, the support playbook, the sales battlecard, the partner training guide, and the implementation steps are all generated directly from the same underlying context. Now you can build a definitive source-of-truth and cascade it automatically through the organization, repackaged using the templates every team has already refined through hundreds of launches. The product launch that used to take months of coordination across dozens of teams can compress into something close to instantaneous.No more coordination tax. AI isn’t perfect today. It hallucinates. It deviates and meanders. You still need humans in the loop to double check the work, to refine the prompts and workflows based on feedback, and to maintain the templates. You just need way, way fewer of them. And this is the worst AI will ever be.AI is transforming innovation and marketing too, but the difference is that context carriers are being automated away while innovators and marketers are being augmented. An engineer with AI can ship more in a week than they used to in a month. A star sales rep can prep for twice as many meetings, personalize outreach at scale, and enter every call better prepared in minutes than they used to be after hours of research.When a business accelerates their pace of innovation or their sales velocity, they never ask themselves: “can we hire fewer engineers or sales reps?” They ask: “how do we ship more product and put our best people in front of more customers?” But when AI makes knowledge documentation, enablement, implementation, or support more efficient, they immediately think: “I can probably do this function with 1/4 the headcount.” And marketing, in the broadest Druckerian sense, is about humans building relationships with buyers. Outside of truly touchless, high-volume B2C, basically everything in business comes down to relationships. AI can help sales reps cover more ground and walk into every call more prepared. AI cannot sit across the table (or Zoom) from a prospective buyer and build a genuine connection that makes dollars move bank accounts.In 2025, a solo founder named Maor Shlomo built an AI-powered app development platform (aka a vibe-coding platform) called Base44 as a side hustle. He grew it to 250,000 users and nearly $200,000 a month in profit in six months, bootstrapped, with fewer than 10 employees. Wix acquired it for $80 million in cash. One person built the product and distributed it. Zero context carriers.Now think about what it would take for an incumbent software company to launch a competing product. It would take six months to work its way from idea to the customer through the context chain: documentation, enablement, support training, sales readiness, marketing positioning, partner onboarding.And Base44 isn’t a one-off. There is a whole cohort of AI-native startups who are growing at rates we have never seen before. Consider these two charts from the recent Redpoint Ventures Market Update.First, these AI-native businesses are growing faster than any companies in history:Second, they are growing much more efficiently:Yes, part of this is AI demand being off the charts. Literally. Look at the charts. But demand alone doesn’t explain the efficiency. Stripe and Figma had massive demand, too. A big reason they are growing so quickly and efficiently is they are skipping the context chain entirely. They are building flatter, leaner organizations with more automation to deliver product context as directly as possible to the customer.Incumbents have a much harder problem. Their entire organizational structure is built around context carrying: the roles, the reporting lines, the meeting cadence, the review processes, the templates, the HR requirements around management spans and layers, all the way down to what they incentivize and how they promote. You can't just drop AI into that org chart and expect it to reorganize itself. And the people in those roles aren't going to help the business transform. They are going to drag their feet. Their careers, their identities, and their mortgages are built on being really good at carrying context, and they're going to fight like crazy to protect the status quo.Look at your last two weeks of work and categorize your time into three buckets: 1. How much was innovation? How much time did you spend creating original context that didn’t exist before. Making decisions, building the product, generating the raw material that everything downstream depends on. 2. How much was carrying? How much time did you spend taking context from one place and repackaging it? 3. How much was marketing? How much time did you spend personally getting the product into the hands of an end user?I think many readers, upon reflection, will acknowledge that much more of their role is in category 2 (carrying context) than is comfortable. And the temptation is to say that you’re adding judgement and value bundled with that core context. While that might be true, the risk is that you sound like Tom Smykowski in Office Space telling the Bobs he has people skills.The gap between “I add value to this process” and “I am the irreplaceable source of this” is where a lot of careers are going to get squeezed over the next few years.While AI is automating the context carriers, it is also collapsing the barriers between traditional roles. Roles that used to have clearly delineated responsibilities are being combined and redefined. Think about the traditional product trio of engineer, designer, and PM. Those three roles are merging into something more like a “product builder” who can take an idea from concept to shipped product with dramatically fewer dependencies. Marc Andreessen had a particularly funny bit on Lenny’s podcast about how AI is impacting all three of these roles: “There's a Mexican standoff happening between those [product managers, engineers and designers]. Every coder now believes they can also be a product manager and a designer. Because they have AI. And every product manager thinks they can be a coder and a designer. And then every designer knows they can be a product manager and a coder. So people in each of those roles believe that with AI, they don't need the other two roles anymore.What I think is so fascinating about this Mexican standoff is they're actually all kind of correct. AI is actually now a really good coder. It's actually now a really good designer and it's also a really good product manager.”AI is also causing the same exact role collapse on the go-to-market side. A traditional sales funnel might have five handoffs between marketing, SDRs, AEs, CSMs, and AMs. A seller with AI can now prospect at scale, do their own research, handle follow-up, and run account analysis without handing any of it off. Years ago, way before AI, Snowflake famously killed the CSM job:“We don’t believe companies should have a separate customer success function. The first thing we did when Frank joined Snowflake was we blew up the customer success function. Customer success is not accountable for anything.”Mike Scarpelli, Snowflake CFOIf Snowflake could do it before AI, AI makes it way, way easier today for a Sales Rep to own the relationship for the entire customer lifecycle.The product collapse gives you a single individual who owns the full arc from idea to shipped product. The GTM collapse gives you the “customer owner” who manages the full relationship from first touch to renewal. As the context supply chain collapses, you have three options on how to proceed:Become the context orchestrator (Defense): We’ll need far fewer context carriers, but we’ll need some. This is going to be a knife fight for the remaining slots. If you want one of them, you need to become as AI-pilled as possible, as fast as possible. Know every tool, run every workflow, own the templates and the prompts. This is survivable, but it’s a defensive posture.Move with conviction toward innovation or marketing (Offense): It’s never been easier to build or to sell. AI has collapsed the barriers between roles and you don’t need an army to ship a product. You don’t need four people to each own a piece of the customer relationship. The path from context carrier to innovator or marketer is more accessible than it’s ever been, and the people who move now will have a huge head start on the people who wait. This is the offensive move, and I think it’s the right one for most people.Invent a role that doesn’t exist yet (Wild Card): Every major technology shift creates roles that nobody envisioned. Nobody wanted to be a “social media manager” in 2004, because the role didn’t exist. Right now, the collision of collapsing org structures and agentic AI is going to create jobs that don’t have names yet. If you are a highly-autonomous individual who loves to build, this is your best bet. It’s essential that you also move to a company culture that empowers and celebrates these types of blended, experimental roles. It is no surprise that many startups founded post-2022 are operating this way.I will leave you with this: The goal is to become a token orchestrator before you become a token orchestrated by someone else.Good luck.No posts

※ 著作権に配慮し、引用は冒頭3段落までです。続きは元記事をご覧ください。