The end of the bow tie

AI is a generational disruptor to GTM. Most of what used to work in B2B tech: the funnels, the handoffs, the playbooks, the well-referenced CRO hire, is not in need of optimization. It is invalid. Built on assumptions about buyers, teams and markets that 2026 has already written over.

This is landing at an awkward moment for investors. The era when a clean cap table and a name-brand revenue leader could carry a B2B tech company to its next round is behind us. Returns now come from operating improvement, which means GTM has quietly moved to the center of the value creation conversation for Venture and PE alike. It is the one function that both drives revenue and decides whether the company is defensible enough to command the next valuation. And the GTM leaders who can actually run a motion built for the AI era are in short supply, getting shorter.

You can feel it in the board packs. The same format, the same pipeline slide, the same confident narrative, the same slow drift between the forecast and the actuals. Everyone nods. Nobody quite asks the question underneath. In private, it sounds like this: “is this engine going to work in two years, or are we about to write a follow-on cheque for a motion that is already obsolete?”

The world has changed, and most GTM responses are pointed in the wrong direction

The instinct when something stops working is to work it harder. That is what most GTM teams are doing right now. More activity, more campaigns, more sequences, more content, more tools. The numbers that come out of the far end look similar to the numbers from a year ago, which is usually taken as a sign that at least things are stable. It is not a sign of stability. It is the sound of a motion running in place while the ground underneath it moves.

Buyer behavior has changed more than most GTM leaders have yet admitted to themselves. The first conversation a prospect has about your category is with AI, not with you. They arrive at a shortlist before a single marketing impression has been rendered. They read less, compare more, and trust almost nothing that arrived in their inbox unprompted. When they do reach for human input, they go sideways to peers and communities, not forward to vendors. Product expectations have moved with them. Nobody cares what your software does. They care what outcome it produces, and how likely that outcome is to embarrass them at the next finance review. Whole categories that described themselves as AI-enabled last year are now being asked, politely, whether they have any product-market fit left.

The economics of GTM output have collapsed in parallel. A rep can draft ten personalized sequences in the time it used to take to draft two. A marketer can produce a hundred landing page variants before lunch. This sounds like leverage. It is not. It is the thing the guidelines-writers call slop: volume without context, output without judgment. In the bow tie era, a weak SDR or junior marketer produced forgettable work that was at least contained by their own throughput. AI has removed the throughput limit. The slop now scales. Anyone who has opened LinkedIn on a Tuesday morning has seen the result.

And the rate at which all of this changes is faster than most portfolios can absorb. What qualified someone as AI-native eighteen months ago is now table stakes. The playbook that looked sharp at the start of 2025 is already stale. Most GTM engines are falling a little further behind every quarter, whether or not they hit the number.

The consequence for investors is that GTM has become the single largest source of value destruction and value creation in the portfolio. Not marketing. Not sales. GTM as a whole system: the operating model that turns a product into predictable revenue. It is where capital is being quietly burned right now, and where the strongest returns over the next five years will be earned. Underneath that sits a harder problem. The standard GTM Due Diligence question, “is this motion working?”, is a question about the present. It is not useful for the decision in front of most investors, which is the decision about the next cheque. The question that matters is one step harder: “will this motion still be working in two years, or are we funding a machine that is already obsolete?”

Most diligence processes are not yet built to answer that. Most boards are not yet structured to see the signal that would answer it. Most GTM reports still land in a format designed for a market that has already moved. The answer to where the industry goes from here starts with admitting what is dead.

The traditional B2B SaaS bow tie is dead. Most of the industry is still dressed for the funeral.

The traditional B2B SaaS bow tie did a lot of useful work for fifteen years. It gave the industry a shared map. Awareness, consideration, evaluation, commit, onboard, expand, advocate. Each stage had an owner, a metric, a playbook, and a hire who had run the playbook somewhere else. The whole thing was teachable, repeatable, and forgiving. You could walk into any scale-up in 2019 and recognize the furniture. If growth was soft, you hired someone who had fixed the stage you were stuck on. If the CRO didn’t work out, you hired the next one from the company down the road and the playbook transferred with them.

That portability was the bow tie’s real product. Not the diagram. The fact that you could lift a GTM motion out of one company and press it into another and mostly expect it to work. It is what made GTM hireable, fundable, and boardroom-reportable. It is also, with some distance, what made it fragile.

The bow tie codified three assumptions. Buyers move through stages in sequence. Work gets passed between functions down the funnel. Best practice transfers from one company to the next. Every one is now wrong in the way assumptions go wrong in investing: not obviously, not catastrophically, but enough that any strategy built on them slowly underperforms until everyone wonders why.

Buyers no longer move through stages. They ask an AI, read a peer review, watch a podcast, check one vendor’s site for ten seconds, and arrive at a shortlist. Discovery, evaluation and shortlist now happen inside a single session, often without the vendor ever knowing it happened. The MQL to SQL handoff ritual, solemnly observed in revenue operations reviews across the industry, is now mostly theatre. The status in the CRM describes a stage the buyer left two weeks ago, if they were ever in it. Speaking to board members every week, the symptom that lands first is usually along the lines of “pipeline is down 35% in the last quarter.” The team has missed that the pipeline isn’t down because demand is down. It is down because the bow-tie definition of pipeline no longer corresponds to where buyers actually are.

Work can no longer be passed between silos without something expensive falling out of the handoff. Marketing runs its campaigns, SDRs run their sequences, AEs run their deals, customer success runs its renewals, each team chasing a different target and saying slightly different things to the same buyer. In the bow tie era this was inefficient but survivable. In the AI era the buyer sees through it in minutes. Every lost piece of context is another reason the buyer finishes the call politely and never replies to a follow-up. It is also another reason a CRO walks into a board meeting saying “we’re in deals but we can’t create urgency” or “we aren’t going to hit the commit.” The urgency isn’t missing because the AE didn’t push hard enough. It is missing because the buyer is comparing four vendors who all sound the same and is waiting for someone to give them a reason to choose. The teams getting this right have stopped organizing around functional ownership and started organizing around the customer’s situation, with shared context as the unit of accountability rather than handoff status.

And best practice, the comforting idea that what worked at the last company will work at this one, has stopped transferring. The GTM leader arrives with a playbook that was sharp in 2022, applies it faithfully, and watches it produce weaker numbers than the one it replaced. “The new AEs are taking longer to ramp” is the version of this complaint that lands in the boardroom, but the AEs are not the problem. The playbook they are being onboarded into is. Nobody wants to say out loud that the playbook itself has expired, so the conversation instead becomes about execution, or the team, or the market, or the product. Six months later, the leader leaves. A new one arrives with a different playbook that is also a year out of date. The cycle repeats. The companies that have broken out of it treat playbooks as starting points to be tested and killed quickly, not procedures to be defended for the length of an annual plan.

What replaces the bow tie is not a different funnel shape. It is a different unit of organization. The bow tie organized GTM around stages. The motion that works now organizes GTM around flywheels. Not one flywheel, a set of them, each aimed at a distinct customer context. A flywheel for building brand and awareness among buyers who are still out of market. A flywheel for turning pain-aware prospects into customers. A flywheel for delivering value and growing accounts once they have signed. These do not run in sequence. They run in parallel, continuously, and they reward teams that share context across them rather than pass it down a line.

The difference matters because flywheels compound and funnels do not. A stage terminates. A flywheel keeps turning, and every rotation makes the next one easier. And what makes the rotation possible is context: a persistent, shared understanding of the buyer, the account, the deal, the customer, built up over time and accessible to every function that needs it. In the bow tie era, context lived in people’s heads and was lost on every handoff. In the AI era, context is the asset. It is what makes AI produce signal rather than slop, what lets one flywheel enrich another, what allows a team to act faster without acting dumber. The winners are building context layers, not funnels.

AI GTM maturity is not about the stack. It is about the foundation underneath it.

Walk into any portco in 2026 and you will find AI tooling somewhere in the GTM stack. A meeting notes generator, a research agent, a content co-pilot, an outbound sequencer, something. Ask how it is performing and you will usually get a confident answer. Ask how it fits into the operating model and the answer gets thinner. Ask how it has changed the way the team makes decisions and you will often get silence, or a couple of anecdotes.

The common mistake is treating GTM maturity in the AI era as a tooling question. How much AI, how recent, how well configured. This is backwards. The foundation underneath the tools is what determines whether the tools help or hurt, and the foundation is built out of things the stack page of the board deck never mentions.

1. Whether the operating model can learn, align, and execute faster than the market moves.
2. Whether the strategy is built on a current picture of the buyer or a legacy one.
3. Whether AI is woven through the motion or bolted onto the side of it.
4. Whether the company is structurally dependent on a tiny group of GTM leaders who have successfully made the change.
5. Whether the board is seeing ground truth or a polished version of it.
6. Whether there is an actual plan, with a real timeline, for moving from the GTM the company has to the GTM it needs.

Those six conditions are what durability is made of. They are not a scoring rubric. They are the things that decide whether the GTM motion keeps compounding or quietly stops. And they cut across the usual revenue functional lines, which is part of why most portcos find them uncomfortable to look at squarely.

There is a seventh condition worth naming separately, because most companies are quietly getting it wrong.

1. Whether the company actually knows, with structural clarity, why it wins when it wins.

Not the positioning statement on the website. Not the tagline an agency rebuilt last quarter. Not the deck the consultant left behind after the off-site. The underlying answer to “Do we really understand why we are different in a way that our customers care about?” This is the thing AI has commodified hardest. Anyone can rewrite a positioning deck in a week. Anyone can A/B test a tagline. The deck-of-the-quarter brigade has never been busier, and the work they produce has never mattered less. The companies that win in the AI era are not the ones with the freshest message. They are the ones who have done the harder, slower work of knowing what they are materially better at and why their customers care about it, and then adapting how they say it continuously, rather than picking one version and running with it for a year. Differentiation is not a messaging exercise. It is the foundation underneath the messaging. Most of the work being sold as the former is a substitute for the latter, and the substitute does not compound.

Against that foundation, three tiers show up repeatedly across the portfolios we look at.

The first tier is rare. A handful of companies have cross-functional GTM teams working off a shared context, with AI embedded across the flywheels rather than pinned to individual workflows, and with the board close enough to the motion to guide it rather than audit it. In practice this shows up as a handful of things. They treat context as infrastructure, not an artifact of individual roles. They measure the value of each human interaction, not just the volume of them. They run GTM experiments on cadences measured in days rather than quarterly campaign cycles. And they separate execution from results in their board reporting, so the board sees whether the plan was delivered, not just whether the number was hit. These companies do not always look extraordinary from the outside. Quarterly numbers are often unremarkable for a few cycles, because the compounding takes time to surface. When it does, it is the difference between a company that earns its follow-on on strong terms and one that has to explain why the last cheque is not showing up in the pipeline.

The second tier is where most portcos actually live. Individual GTM functions are using AI to go faster inside their own lane. Marketing is generating more content. SDRs are sending more sequences. AEs are getting prettier account briefs. Activity goes up. Quality stays where it was, or drifts down. The silos that were survivable in the bow tie era now compound against the company, because every function is accelerating independently and the buyer experiences the sum of their misalignment. From the inside it feels like progress. From the outside it looks like a company that is busy without being effective.

The third tier is the most dangerous, because it is often the proudest of its AI adoption. Tools have been bought. Vendors have been onboarded. A quarterly AI update is read into the board minutes. But the foundation, the operating model, the data, the context, the cross-functional alignment, has not been touched. The tooling produces output nobody trusts, acted on by a team that is not aligned, guided by data nobody has cleaned. This is where the slop is loudest, because AI is efficiently generating things nobody should be doing. Teams are working harder than they ever have. The numbers do not trend upwards.

Something worth naming plainly here. AI has removed coordination and execution as constraints. What that exposes, quickly, is whatever was previously compensating for weak judgment. In most GTM teams, something was. When the cost of producing output falls to near zero, the quality of the thinking behind it becomes the only thing that separates signal from noise. Judgment is the new scarce resource. And judgment does not scale just because the tools have.

The single most consistent message from leaders who have actually made it to the first tier is not about technology. It is about the operating change. They talk about unlearning the bow tie habits of mind. About retraining managers to run flywheels rather than stages. About rewiring the reporting cadence so the board sees the motion, not a static snapshot of its output. The tooling, in their telling, was the easy part. The hard part was the thinking and the operating. We find this consistent to a degree that is almost monotonous. The companies that win do not start with the stack. They start with the people, the operations, and the evaluation. The stack follows.

Which raises the obvious question. If the foundation is what matters and most portfolios are not built on one, what is the cost of that gap over the next two years?

Durability compounds. Fragility compounds faster than anyone expects.

The gap between tier-one GTM and tier-three GTM does not stay where it is. It widens, quarter by quarter, in a way that is almost invisible from inside the company and becomes unignorable from outside it.

Durable GTM compounds for the same reason as any learning system compounds. Every campaign produces data that sharpens the next. Every customer conversation feeds the positioning. Every decision in the operating cadence improves the quality of the next decision, because the context that informed it is now in the system rather than in the head of whoever ran the meeting. A team operating this way gets better at its job as a condition of doing its job. The rotation of the flywheel is itself the learning.

Three layers have to move together for the compounding to start: operations, leadership, and board visibility. Most portfolios have none of the three fully installed. A handful have one. Almost none have all three. This is what we mean when we say there is an operating model gap.

The investor asymmetry hidden in all of this is what makes the next two years unusual. A portfolio approach to durable GTM is not just additive across companies, it compounds across them. Patterns from one portco sharpen the diligence on the next. The operating model that works in one vertical, stripped to its structural elements, transfers into another. The GTM leaders developed inside the system become the hires for the companies that need them. This is the kind of structural edge that quietly separates funds that post strong DPI from funds that disappoint at exit. The window to build it is open now because most investors are still treating GTM as each portco’s separate problem. Once that stops being true, the edge closes.

Which brings us to the claim we have been circling. The companies that build durable GTM in the next twelve months will earn their follow-on rounds, or their exits, on structurally different terms than the ones that do not. We believe this gap is about to become a diligence discount. Not priced in yet, but coming. The first serious buyer to ask “what it would cost to rebuild this company’s GTM from the foundations up, and how much growth would be lost while they do it?”, will get a different answer than they expect, and they will start pricing that answer into the bid. The rebuild itself is expensive. The twelve to eighteen months of soft numbers while it is underway is more expensive. Once one fund starts pricing in both, the others cannot afford not to. The discount, once it shows up, will move fast.

That is our thesis in one sentence. GTM durability is about to be a pricing input. The portfolios that invest in it now will find themselves on the right side of that shift. The portfolios that do not will be on the other side, wondering why the diligence conversations got harder and the valuations got softer.

Crossing the chasm starts with giving up on the hero

Recognizing the gap is the straightforward part. Closing it is where most portfolios run into trouble, because the usual closing move no longer works.

The usual closing move is to find a better CRO, CMO, or VP. Hire someone who has done it before, someone with the right logos on their Resume and the right references from someone the partner trusts, and let them rebuild the GTM motion. This instinct was entirely reasonable three years ago. It is no longer sufficient, because the market has stopped supplying the people. The pool of GTM leaders who have actually built an AI era motion around context, flywheels, cross-functional alignment and embedded AI is small. The pool is being bid up across every portfolio at once. The best people already sit on more boards than they can meaningfully operate, and their bandwidth for any one engagement is thinner than the LinkedIn profile suggests. Whatever the market looked like the last time you hired, it is tighter now, and the definition of AI-native that satisfied a board twelve months ago does not satisfy one today.

This is the chasm. Geoffrey Moore’s original framing holds up well here. The early adopters of AI era GTM got themselves across on founder intuition, one or two unusually capable hires, and a lot of rebuilding in flight. That move does not scale. The mainstream crossing requires treating the operating system itself as the asset and the leader as the operator of it. This flips the dependency that has quietly broken most portcos. The system does not collapse when the leader leaves. New leaders learn the system, not invent their own. Change management becomes possible because the end state is specified rather than discovered.

This is the harder move politically, because it asks the CEO and the CRO to admit that the right answer is not another hire. It asks the board to stop looking for the next savior. It asks everyone in the room to accept that the company’s GTM is a system they have to build deliberately, not a talent problem they can recruit their way out of. In our experience this is the moment most companies flinch. The savior hypothesis is comforting. It promises a clean handoff and an inspiring narrative for the next board meeting. The system-installation hypothesis promises months of uncomfortable work and a GTM leadership team that will need to unlearn habits they have been rewarded for for years.

The boards that get their companies across share one structural feature. They are inside the work rather than outside it. They see GTM operating health on something closer to a monthly cadence than a quarterly one. They see a forward-looking view of the motion, not a retrospective tour of the outputs. They understand which parts of the operating system are learning and which are stuck. This is a meaningful shift from the governance model most boards have operated for the last decade, where a few metrics were included in the monthly board report, with more detail provided for a quarterly board meeting. In the AI era, a quarter is a long time for drift to compound. Most boards we talk to might be looking at numbers monthly, but are still running the quarterly cadence because it is what everyone is used to, not because it fits the environment.

What this requires of the GTM leaders themselves is worth naming. The job has changed. It is no longer running a proven playbook with confidence. It is running a learning motion with rigor, building and killing experiments on faster cadences than anyone is comfortable with. It is protecting the team from the slop that AI has made so cheap to produce. It is closer to operating a trading desk than running a factory.

All of which points to a different starting move. Before the next hire, before the next tool, before the next agency retainer. Audit honestly. Look at the motion the company is currently running, the operating model underneath it, the board’s line of sight to the ground truth, and the plan, if there is one, for the motion the company will need to be running in two years. If any of those four looks thin, the problem is not a pipeline problem. It is a foundation problem.

Ten questions every board should be able to answer

The cleanest diagnostic for where a company sits against the argument above is also the simplest. Pick the next board meeting on the calendar. Walk in with these ten questions. Ask them plainly and listen to what happens.

What you are listening for is confidence. The kind that comes from a team that has already thought about the question, worked to find the answer and can defend it, not the kind that comes from a team pattern-matching to what the board probably wants to hear. Confident answers are short and specific. Unconfident answers are long and qualified. The difference is usually obvious inside thirty seconds.

The ten questions, in the order we would ask them:

1. How has AI changed our buyers’ behavior in the last twelve months, and what are we doing differently because of it?
2. Who will be our ICP and economic buyer in two years, and do we know what they care about today?
3. Is our category durable in the AI era, or is our positioning built for a market that is already moving past us?
4. Do we really understand why we are different in a way that our customers care about? Or do we have a positioning deck the agency rebuilt last quarter?
5. Do we have a GTM operating model built for the speed of the AI era, or are we mostly relying on established playbooks?
6. Is AI embedded across our GTM motion, or bolted onto individual functions where it mostly produces more of the same output?
7. What is our single source of truth for GTM performance, and who sees it in something close to real time?
8. Are we developing GTM leaders for the motion we will need in two years, or hiring for the motion that worked two years ago?
9. What GTM activity have we deliberately stopped in the last six months because it no longer works, and what did we learn from stopping it?
10. If our GTM leaders left tomorrow, would the operating system survive, or would we be starting again?

The questions are deliberately uncomfortable. They are written to find the places where a confident narrative has been papering over a structural gap.

Three bands, loosely:

Eight to ten confident answers is a durable GTM motion. The priority from here is compounding. Make sure the system keeps learning faster than the market moves, and make sure the board stays close enough to see when it stops.

Five to seven confident answers is a partially durable motion with specific gaps. The risk is that the gaps compound. The highest-leverage move is almost always fixing the weakest of the three layers: operations, leadership, or board visibility, rather than adding more activity on top.

Fewer than five confident answers means the foundation is not yet in place. Tooling and tactics will not close the gap. The work starts with how the team operates, how the board sees, and how decisions get made. Anything layered on top of a missing foundation is effort without impact, which is the most expensive kind.

What to do now

Three moves, in order, regardless of where the scorecard lands.

Audit honestly. Work through the ten questions with the team and the board together, not separately, and watch where the confidence drops.

The gap is almost never where the numbers suggest it is.

Start with the operating model, not the stack. The AI tool the company buys next quarter will either amplify a strong motion or accelerate a weak one. That decision has already been made by the operating model underneath. Fix that first.

Shorten the board feedback loop. Quarterly visibility was built for an environment that no longer exists. Monthly is the new floor. Weekly, on operating health, is where the tier-one companies already are.

The companies that do this work in 2026 will not look different on the surface for a while. They will look different in the follow-on round.

Demand Karma helps VC and PE-backed B2B tech companies between $5M and $25M in revenue build Agentic GTM for the AI era. We run an Agentic GTM Audit. A focused review that audits the current levels of Agentic GTM maturity, by team and aspect. Providing a clear view and prioritised actions for investors, boards, and CEOs. If any of this resonated, the next conversation is worth having.

Edwin Abl & Josh Morse, Demand Karma, 2026