The Custodian · An Investor's Story

Close the Loop — What the Investigation Revealed

Custodian You went back to investigate the conditions of reproducibility. What did you find?

James The founder had one customer who came through a cold channel — found the product without a warm introduction, went through a standard procurement process, implemented without the founder present. Outcomes were strong. Usage was deeper than the network-introduced customers, actually, because the operations team had no alternative to lean on — they had to make the product work. That was the data point I needed. The product holds outside the protected conditions. I led the round.

Custodian What changed in how you structured the investment?

James Milestones tied to unselected customer outcomes — not revenue growth alone, but expansion rates and usage depth in customers who didn't come through the founder's network. And I asked for early visibility into what the deployment looks like without founder involvement. Not because I distrust the founder. Because that's the layer I need to be reading going forward.

Custodian You invested in a transition, not a demonstration. That changes the monitoring posture as much as it changes the entry decision.

James Yes. And I can feel the difference. When I look at the board reporting now I'm reading it differently — not just whether revenue is growing but whether the conditions of growth are changing in the direction of full market reproducibility or staying inside the protected zone.

Something has shifted. This is no longer a single experience. It's becoming a different way of reading every position in the portfolio.

The Second Decision — A Follow-On Commitment

Custodian What's the next major commitment you're moving toward?

James A follow-on decision in a Series B company — one of our best-performing portfolio positions. The company is in the HR technology space. They were early to the thesis that performance management as a category was built around a structural constraint — annual review cycles — that continuous feedback infrastructure was making removable. Revenue has compounded at over 80% annually for three years. Enterprise adoption is strong. The Series B terms are attractive. The obvious move is to follow on at our pro-rata.

Custodian Describe what the company looks like from inside the organisation right now. Not the metrics. The organisation.

James The founder is under significant pressure. They've hired a professional management team to scale beyond the early stage, and there's real tension between the original category-creation logic and what the enterprise customers are asking for. Some of the new VP-level hires come from traditional HR software companies and are pushing for feature parity with established platforms — annual goal-setting workflows, calibration tools, the standard functionality the existing category provides. The founder is navigating that internally while managing a board that is primarily focused on the revenue metrics.

Custodian How long has that tension been visible?

James Probably six months. The revenue metrics haven't reflected it yet. The product roadmap has started to reflect it — there are features in the last two releases that are clearly designed for the old category's buyers rather than the new one.

Custodian The revenue metrics continue advancing. The structural position underneath is moving. Those are two different layers — and you've just described reading the right one.

James stops. He has been framing the follow-on as a straightforward decision supported by strong metrics. The framing just collapsed into something more complicated.

Conditions examined — before follow-on commitment

The VP-level hires from traditional HR software companies are not simply adding execution capacity. They are importing the old category's logic into the organisation's daily decision-making. Every feature request they champion, every customer conversation they lead, every roadmap prioritisation they influence is filtered through an understanding of what HR technology is supposed to do — built from the category the company was created to displace. The organisation is accumulating institutional pressure toward the old governing logic from inside its own management layer.
The product roadmap has already begun reflecting this drift — features designed for annual goal-setting workflows and calibration tools are the old category's architecture appearing inside the new one. The founder hasn't lost the category-creation thesis. The organisation is progressively building infrastructure that embeds the old thesis alongside it. In twelve months the product will be structurally ambiguous — neither the old category nor the new one, but something that satisfies neither fully.
The board's focus on revenue metrics during this period is providing cover for the structural drift. Because the metrics are advancing, there is no visible signal that the structural position is weakening. The founder faces sustained internal pressure toward the old logic with no instrument that makes the drift visible before the metrics reflect it. By the time the revenue metrics catch up to the structural deterioration, the follow-on decision will already be made and the window for meaningful correction may have narrowed significantly.

Custodian Were any of those visible before now?

James The first two I could see the surface of but hadn't named as structural drift. I was treating the management team tension as a normal scaling challenge — founders always struggle with the transition from early to growth stage. I hadn't connected it to the governing logic of the category. The third one is the one that's most uncomfortable. The metrics I've been using to monitor this position are providing false confidence at exactly the moment when the structural position most needs to be visible.

Custodian What do you do?

James I don't follow on automatically. I have a different conversation with the founder first — not about the revenue metrics but about the product roadmap decisions and what's driving them. I want to understand whether the founder still holds the category-creation thesis with the same structural precision they held it at Series A, or whether the pressure has started producing accommodation that the metrics aren't yet reflecting. That conversation will tell me more about this position than the revenue chart.

Custodian You just changed the follow-on diagnostic from revenue-based to condition-based. The same shift you made at the entry decision, applied to the position management decision.

James The same underlying question. Not how strong are the results. But what is happening to the structural conditions that produced them.

"The metrics were advancing. The structural position underneath was moving. Those are two different layers."

Naming What's Underneath

Custodian Notice what's happened across both decisions. The supply chain pre-seed and the HR technology follow-on. The invisible conditions aren't random. They're all generated by the same structural mechanism — the gap between what the metrics measure and what actually determines whether the transition succeeds. At entry, the gap hides whether the demonstration is genuine. During the position, the gap hides whether the structural logic is holding.

James Two different failure modes. The first is selection — backing a demonstration rather than a transition. The second is holding — watching a genuine transition lose structural integrity before settlement arrives.

Custodian Exactly. And most investment infrastructure addresses only the first. The entire apparatus of due diligence, reference calls, competitive analysis, and pattern recognition is oriented toward selection. It asks: is this a real opportunity? It does not ask: will the structural logic hold through the transition window?

James Because there's no instrument for the second question. You monitor the metrics and hope the structure is holding. When it stops holding — when the founder drifts, when the organisation accumulates the old logic, when the product starts serving the wrong category — the metrics don't reflect it until it's already happened. By then you're managing a recovery, not a transition.

Custodian Which is why the portfolio company you described at the start — the one whose category never separated — wasn't a selection error alone. It may also have been a holding failure. The structural logic may have started drifting before the metrics reflected it, and neither you nor the founder had visibility into that drift while it was still actionable.

James goes quiet. The RPA company he backed three years ago is no longer just a selection problem. It's become a case study in both failure modes operating simultaneously — a demonstration that was never fully tested against the full market, inside an organisation that was progressively losing structural fidelity under pressure. And he had no instrument that could see either layer while there was still time to act.

Why the Gap Keeps Producing the Same Blindness

James opens before the Custodian does.

James I've been sitting with the two failure modes. Selection failure and holding failure. I'm using the conditions-based diagnostic now for entry decisions and it's already producing different conversations with founders. But I can't yet answer the deeper question. Why does this keep happening to sophisticated investors who understand the category creation thesis? Why does the gap between the metrics layer and the structural layer persist even when the investor knows it exists? What's the mechanism that keeps producing this blindness?

Custodian That's the right question. And the answer has two parts — one about the field you're evaluating, and one about the position you're evaluating it from.

Custodian The field part first. Category creation investing generates false positives structurally — not because investors are careless or founders are deceptive, but because the transition window produces genuine and simulated demonstrations that look identical from the outside. Both produce real revenue, real customers, real institutional endorsement. The difference is invisible in the metrics. It lives in the conditions under which those outcomes were produced. Pattern recognition — the accumulated heuristics of what successful category creators look like — cannot separate genuine from simulated during the transition window. It can only confirm the separation after compression has already arrived and the investment decisions are already made.

James Which is why the power law persists. Not because genuine category leaders are naturally rare. Because the diagnostic field is operating on pattern recognition — which gives the right answer too late.

Custodian That's the field part. Now the position part. The investor is not outside the field they're evaluating. The same pressures that distort founder cognition — survival pressure, social proof, narrative coherence, institutional momentum — operate on investors too. And they operate most powerfully at exactly the moment when false positives are hardest to detect: when the narrative is most compelling, when confirmation is arriving from every direction, when the cost of missing out feels higher than the cost of being wrong.

James The conditions-based diagnostic has to be applied before those pressures consolidate. Once you're inside an active process with a compelling narrative and social confirmation from co-investors, you're applying the diagnostic inside the simulation field.

Custodian Exactly. The diagnostic earns its value when the narrative is most compelling — which is when it's hardest to apply and most expensive to skip.

"The investor is not outside the field they're evaluating. The same pressures that produce false positives in founders operate on investors — most powerfully when the narrative is most compelling."

The Third Decision — He Brings It Himself

James I want to run something through it. We're evaluating a company in the AI-enabled legal technology space. The narrative is exceptionally compelling — they're making a genuine structural argument about why the billable hour model is becoming incoherent as AI reduces research time to near-zero. The founding team is credible. Three top-tier firms are competing for the round. Every credible investor in our network who has seen the company has backed or is considering backing it. The social confirmation is as strong as I've seen in this fund cycle. And I can feel the conditions-based diagnostic getting harder to apply under that weight.

Custodian That's exactly the moment the diagnostic matters most. Run it. What do you know about the conditions of the demonstrated outcomes?

James The early adopters are boutique firms and in-house legal teams at technology companies — buyers who were already frustrated with the billable hour model and actively looking for alternatives. The founding team has deep relationships in this specific segment. The deployments have been heavily supported. And the pricing is structured as a subscription well below what the product would need to be at the margins the company is projecting.

Custodian Now apply the three conditions. Can the outcomes be reproduced without the founding team's direct involvement? Were the customers selected from the full market population? Did the outcomes hold inside organisations still carrying the full weight of the established billing logic?

James No, probably not yet, to the first. No to the second — these are early adopters already predisposed to the new model. And no to the third — every organisation they've deployed in was actively seeking to move away from the billable hour before they encountered the product. The full weight of the established logic includes partnership structures, client billing expectations, regulatory frameworks, and law school training that assumes the billable hour as the fundamental unit of legal work. None of the deployments were inside organisations carrying that full weight.

Custodian What does that tell you about the investment?

James It tells me the structural thesis may be right and the demonstration may not yet answer whether the transition is real. The billable hour model may genuinely be becoming incoherent as AI reduces research time. That's a structural claim I find compelling. What I don't yet have evidence for is whether the product's demonstrated superiority holds inside a traditional firm — one where partner compensation is tied to billable hours, where clients have built their legal budgets around hourly rates, where the entire evaluation culture assumes the old model is rational. That's the test the existing deployments haven't run.

Custodian And the three competing top-tier firms who are all moving toward the round — what does their simultaneous presence tell you?

James That the narrative is compelling and the social confirmation is strong. Which are exactly the conditions under which the diagnostic is hardest to apply and most necessary. Consensus among sophisticated investors is not evidence of structural position. It's evidence that the narrative is coherent and the pattern recognition is firing. Those are different things.

Conditions examined — by James, before commitment

The demonstrated outcomes were produced inside organisations already predisposed to the new model — boutique firms and technology company legal teams where leadership was actively seeking an alternative to the billable hour. These are the earliest of early adopters. The full market includes traditional partnership-structured firms where compensation, client relationships, and institutional identity are organised around the model the product claims to displace. The distance between the conditions of demonstrated superiority and the conditions of the full market is very large.
The competitive dynamic among top-tier investors is producing urgency that compresses the time available for conditions-based investigation. The social confirmation arriving from every credible direction is creating exactly the cognitive environment in which the metrics layer obscures the structural layer. The pressure to decide quickly is the pressure to decide on pattern recognition rather than causal diagnosis.
The structural thesis about the billable hour may be correct without the transition being ready to happen at scale. A constraint becoming theoretically removable and a market beginning to actually remove it are different stages. The existing deployments don't yet demonstrate that organisations carrying the full weight of the established model — partnership structures, client billing conventions, training pipelines, regulatory frameworks — will move in the way the thesis requires. The thesis needs a different kind of evidence than the current deployments have produced.

James I'm not passing. The structural thesis is too compelling and the founding team is too strong. But I'm not leading, and I'm not deciding on the timeline the competitive dynamic is trying to impose. I want two things: evidence of one deployment inside a traditional partnership-structured firm, and a conversation with the founder about what they believe will happen when the product encounters partner compensation structures that actively resist the new model. How they answer that question will tell me more about whether this is a demonstration or a transition than anything else I've seen.

Custodian Third decision. Third conditions-based restructuring before commitment. You ran this one entirely yourself — against the strongest social pressure you've described.

James Because I could feel the pressure. And feeling it was the signal that the diagnostic was most necessary. The moment I'm most tempted to skip it is the moment it has most value.

"The moment I'm most tempted to skip the diagnostic is the moment it has most value."

Three decisions restructured before commitment. Two failure modes named and separated. An investor who arrived describing a portfolio company he couldn't fully explain and leaves holding the mechanism that produced it — and the diagnostic that would have surfaced it earlier.

He doesn't carry the Custodian as a diagnostic tool he applies when something feels uncertain. He understands what it's doing. Which means the next investment committee where he brings it into the deliberation will sound different to his partners — not more cautious, but more precise about which layer the evidence is actually answering. The Custodian doesn't vote. It holds the conditions-based question through the deliberation — the layer the committee's pattern recognition doesn't reach — without the social pressures that operate on members who do.

The Right Position. The Wrong Assumption About What Holds It.