Relationships carried B2B for thirty years. They are no longer enough. Not because relationships stopped mattering — but because the specific thing relationships used to provide has been quietly commoditised by a technology sitting on every buyer's laptop. This is the pillar of a five-part thesis on what changed, what replaces it, and what that demands of any CEO — product or services — still running the old motion.
If you have built and scaled a B2B company — whether you sell software, infrastructure, consulting, systems integration, managed services, industrial equipment, professional services, or any combination — you already know the motion. A senior seller spends years building trust with a decision-maker. That relationship becomes the primary channel through which work flows. When the buyer has a problem, they call the seller before they Google. When budget opens up, the seller hears about it before the RFP goes out. When a competitor tries to displace you, the relationship is the moat.
This worked because relationships solved a specific buyer problem. Buyers at enterprise scale were making six and seven-figure commitments based on incomplete information. They did not know which vendors could actually deliver, which consultants would be good, which architectures would scale, which pricing was fair, which products were solid in production versus demo-ware. The market was opaque to them. A trusted seller was the shortcut: this person has not screwed me before, and their firm has done similar work, so this is the lower-risk path.
Relationships in enterprise B2B were never primarily about friendship or charisma. They were an information asymmetry arbitrage. The seller had deep context on product fit, delivery risk, pricing flexibility, internal politics, and war stories from similar engagements. The buyer had limited ability to independently verify any of it. Trust was the bridge across that gap.
The firms that won at scale built institutional versions of this. For product companies: senior AEs with deep expertise, executive sponsor programs, customer advisory boards, QBRs dressed up as strategic alignment but really functioning as relationship maintenance. For services companies: partner-led sales cultures, account farming by senior practitioners, entire growth strategies predicated on kept-warm senior relationships. Different mechanics, same underlying economics.
For a long time, this was correct. It is how most B2B categories got built. The problem is not that this motion is dead. The problem is that the foundation it rested on — information asymmetry — has been quietly removed from underneath it, and most firms have not yet operationally absorbed what that means.
This is the part that gets mischaracterized most often, so it is worth being careful about. Buyers have had Google for twenty-five years. Buyers have had trade press, analyst reports, peer communities, Gartner, G2, Capterra, and SaaS comparison sites for a decade. If the argument were simply "buyers have access to more information," it would have been true for a generation, and the relationship-led motion should have broken down in 2010, not now.
It did not, because search-era information and LLM-era information are two structurally different things. Search returned a menu. The buyer still had to read ten sources, reconcile their contradictions, weigh their credibility, filter out the SEO content, decode the vendor-sponsored positioning, and synthesize a point of view on their own. That synthesis was hard, slow, and unevenly distributed across buying organisations. Most buyers did not have the time or the analytical bandwidth to do it well — which is exactly why the trusted seller remained valuable. The seller was the synthesis shortcut.
Search handed the buyer options. The LLM hands the buyer an opinion. That is the difference, and it is the whole difference. A buyer who now asks an LLM "should we use Vendor A or Vendor B for this?" does not receive a list of links. They receive a reasoned recommendation, with a stated rationale, which they can then interrogate — push back on, refine, challenge with their own constraints, argue into a position that matches their own situation.
What the LLM actually delivers, and what search never could, is a dialogic synthesis. The buyer is no longer a passive reader of ten articles. They are in conversation with a tireless, well-informed analyst who will take their half-formed intuitions, argue with them, sharpen them, and produce a defensible point of view calibrated to their specific situation. This happens before any seller is in the room.
And here is the second-order effect that most vendor sales motions have not absorbed: if the buyer does not arrive with a point of view, the LLM's point of view becomes theirs by default. The buyer who enters a category evaluation without strong priors asks a few questions, accepts the synthesized framing, and walks into the first vendor call already anchored — not on their own analysis, and not on the vendor's positioning, but on whatever framing their LLM produced three days ago. That framing may have named two competitors, favored a particular architecture, dismissed a pricing model, and set the scoring criteria. The vendor meeting the buyer for the first time is now negotiating against a POV the buyer themselves did not consciously construct.
A buyer evaluating a category — a SaaS platform, a managed services provider, an ERP partner, a data infrastructure vendor, an industrial controls system — can in thirty minutes of LLM-assisted research produce a sharper comparative framing than their own procurement team could have produced in six weeks two years ago. They can identify vendors with relevant case studies. They can pressure-test pricing against benchmarks. They can draft their own RFP. They can score vendors against criteria they and their LLM co-authored. None of that involves a seller.
The relationship still opens the door. It no longer closes the deal. And for the first time in a generation, that is two different things.
The seller who used to arrive at the first meeting carrying information the buyer could not easily get is now arriving at a meeting where the buyer has already been talked into a position by a system that had no interest in the outcome, no bias for any vendor, and considerably more patience than any AE. The conversation the seller expects to have — "let me walk you through how we approach this" — is a conversation the buyer already had with their LLM yesterday, faster, more rigorously, and without anyone trying to sell them anything. If the seller's opening move is to educate the buyer, the buyer has already been educated by a system whose synthesis the buyer now half-trusts as their own.
This is not a complaint about AI. This is a structural observation. The specific thing relationships used to monetize — being the synthesis shortcut across an information landscape too dense for the buyer to decode alone — is the thing AI has most efficiently erased. The synthesis is now free, dialogic, and on-demand.
It is worth being precise about what has changed and what has not. The LLM is now better than most sellers at describing options generically, comparing vendor capabilities at a category level, summarizing typical risks, drafting technical specifications, building initial business cases, and answering factual questions about publicly available product or service information.
The LLM is not better than a good seller at: knowing that the CIO at a specific account just hired a new VP whose mandate is a directional shift, and that this changes the buying committee in a way that has a 90-day window. Knowing that the procurement lead at that account got burned on a previous vendor and now requires references before considering new firms. Knowing that a key stakeholder mentioned, on a podcast three weeks ago, a specific initiative that matches exactly what your firm just delivered or shipped.
The LLM can synthesize what is publicly knowable. It cannot synthesize what is happening inside a specific account this week. That kind of knowledge — specific, timely, account-and-person-level context — is the new asymmetry. It is what the buyer cannot get from their LLM. It is what a seller could provide. And it is exactly what most firms, product and services alike, are failing to operationalize, because the systems that would produce it do not exist in their current go-to-market motion.
Start from the buyer's position, not the seller's. The buyer now has a synthesized point of view they arrived at in conversation with their LLM. That point of view was produced from public information — what is on vendor websites, in analyst reports, in reviews, in press releases, in podcasts, in earnings calls. It is a reasonable synthesis of everything the internet knows about the category.
There are exactly two things the LLM could not contribute to that synthesis, because the LLM does not have access to them.
The first is what just happened inside this specific account. A new decision-maker joining last month. A regulatory change the buyer's compliance team started worrying about in the last quarter. A funded initiative that is not yet public. A procurement burn on a previous vendor. An expansion the product team is quietly planning. These are real, consequential facts about the buyer's situation that no LLM has synthesized, because no LLM was told.
The second is what a seller chooses to do with that knowledge, right now, before the buying window closes. Information that is specific to this account but arrives six weeks after the signal is a postmortem. Information that is timely but generic is spam. What is scarce — and therefore what the seller can still charge for — is the combination of both.
Everything else in a vendor's go-to-market motion — category positioning, product education, competitive framing, even most "value propositions" — is now commodity. The only remaining non-commodity seller output is specific, timely action on what is happening inside this particular account. That output has two components that must arrive together: speed and precision.
Speed is the window between a signal appearing in an account and the seller acting on it. A new decision-maker joining. A new initiative getting funded. A competitor announcing a win. A regulatory change affecting the buyer's industry. A post from a senior stakeholder signaling a priority shift. A trigger in the buyer's own earnings call. An expansion signal in an existing account. A churn risk surfaced by product usage data.
Speed matters now in a way it did not five years ago because the buyer's point of view hardens faster. When a stakeholder forms their POV in a forty-minute LLM session, the window in which that POV is open to new inputs is days, not quarters. The seller who surfaces a relevant, account-specific fact while the buyer is still actively refining their view gets added to the framing. The seller who shows up six weeks later is arguing against a POV that has already calcified — and arguing against your LLM-formed view is, for most humans, more uncomfortable than arguing against a vendor's pitch.
In the old motion, signal latency was absorbed by relationship depth. A senior seller could get a meeting regardless of timing because the relationship carried the conversation. In the new motion, that relationship still gets the meeting — but the meeting now competes with a view the buyer has already taken. Latency no longer costs a pipeline metric. It costs the right to shape the POV.
Precision is the specificity of what the seller does with the signal. A generic nurture email to a new decision-maker is not precision. A tailored point-of-view on a specific initiative, referencing a specific analogous customer outcome, naming a specific stakeholder's priority, sent at a specific moment in the buying cycle — that is precision.
Precision is what makes the seller still valuable when the LLM has commoditized generic expertise. The LLM can produce generic content at near-zero cost — and increasingly, the buyer's own LLM is doing so for them, pre-empting the vendor's outreach. What the LLM cannot produce, because it does not have access to the context, is the specific combination of this signal, this account, this stakeholder, this moment, this POV. That combination is the new sales artefact. And producing it at volume, consistently, across a seller team, is now the operational capability that separates firms pulling ahead from firms being outpaced.
Speed without precision is spam. Precision without speed is a postmortem. The firms winning do both — not because their sellers are exceptional, but because they have built a system that makes both the default, not the heroic exception.
Everything above applies equally to a SaaS founder watching incumbent platforms get circled by lighter, AI-native alternatives, and to a services firm watching buyers arrive at the first call with a better opinion of the market than the AE. The information-asymmetry collapse is the same. The speed-with-precision response is the same at the abstract level.
But the specific operational implications diverge sharply — and this is where most CEOs stall, because generic AI-shift advice is useless at the point of execution. A product-company CEO needs to think about whether their product is getting commoditized, whether customers can now build a lighter in-house version, where stickiness and expansion come from when feature parity is no longer a defensible moat, and what happens if compliance-as-a-service erases one of their remaining advantages. A services-firm CEO needs to think about what they are actually selling — people, projects, or outcomes — and whether outcome-based pricing is a fit or a trap. Both need to think about whether their marketing and distribution motions are still reaching buyers who now research via LLMs, not search engines. And both need to rethink what relationship selling means when the relationship is no longer the information bridge.
Those are four distinct operational arguments. Each deserves its own article. Each is genuinely different for product versus services. Bundling them into one generic "what to do about AI" essay is how most of this category's thinking fails — the arguments collapse into platitudes because they are trying to speak to everyone at once.
The rest of this series takes each one specifically.
Read them in order if you want the full argument. Jump directly to the one that matches your context if you want to get to the operational question faster.
Every major shift in enterprise buying has a transition window — a period during which the firms that absorb the new model pull meaningfully ahead of the firms that defer the decision. Coming out of that window, the gap is very hard to close, because the operational debt compounds: the firms that absorbed the shift early now have multiple quarters of historical data, calibrated motions, and improving economics. The firms that deferred are starting from zero at a moment when their buyers have already adjusted.
We are somewhere in the middle of that window right now. The firms that started by treating the AI shift as an operational priority are currently pulling ahead on retention, expansion velocity, cost of customer acquisition, and — most visibly to investors — the ability to describe their go-to-market engine in specifics rather than narrative. The firms that are still treating this as next year's problem are the firms that will spend the next year explaining why growth decelerated while a competitor half their size quietly took market share in accounts and markets they thought they owned.
This transition is not primarily about adopting a tool.
It is about whether the firm's go-to-market motion is instrumented enough to be improvable. Tools are downstream of that decision. Firms that have made the operational decision choose some tool and move forward. Firms that have not will evaluate tools indefinitely and move nowhere. The decision is not technical. It is a CEO-level commitment to run go-to-market as an operating system instead of a trust-based function.
The specific question worth sitting with — before the next board cycle, before the next planning review, before the next leadership one-on-one:
If your buyer arrives at the first call with a point of view on your category — a POV they either constructed or accepted from an LLM — what, specifically, is your team adding that can still shift it?
If the honest answer involves the relationship, the responsiveness, the specific account context your team brings, or the timeliness with which they act — you are on the right track. But you are describing something that has to be systematically produced now, not held in a few people's heads. Producing it reliably is what the next four articles work through — operationally, by category.
Part 2 takes the product-company case. Part 3 takes services. Part 4 is the marketing and distribution response. Part 5 is the synthesis — what relationship selling actually becomes in the era when the relationship is no longer the information bridge. Read them in order, or jump to the one that matches your context.