# The task nobody has time for: thesis invalidation as a monitored condition. Every investment memo says why. Almost none are reopened to check whether the why still holds. This is what changes when a memo stops being a document and becomes a monitor: how the conditions get written, what happens when one trips, and why the task needed machinery no manual process could supply. Published: 2026-08-14 Updated: 2026-08-15 ## In brief - The memo dies at the moment of conviction, and the cause is **arithmetic**: a continuous task with no deadline loses to every deadline the firm has. - The cost shows up as **positions that outlived their premise**, exited on price action instead of thesis. - The fix is small at writing time: **stated failure conditions become monitored items** on the same lifecycle as every other obligation. - A tripped condition arrives specific, so the team **revises, acts, or writes down why the change is survivable**, on the record. ## The premise :: The memo dies at the moment of conviction. The investment memo is the best artifact most firms produce. It states the thesis, the sizing, the risks, and, explicitly or otherwise, the conditions under which the position should not exist. Then the trade goes on, the memo goes into the drive, and it is never opened again. This is not negligence. It is arithmetic. Reopening every memo against current conditions, position by position, week after week, is a continuous task with no natural deadline, no external chaser and no visible cost when skipped. Tasks with that shape lose to every deadline the firm has. Always. ## The cost :: What it costs not to do it. The failure mode is quiet. A position is held for reasons that stopped being true. The margin story faded. The competitor shipped. The regulatory posture turned. The customer concentration the memo flagged as tolerable became the whole book. Nobody decided to keep holding the position; the decision simply never came up for review, because nothing forced it to. The eventual exit then happens on price action instead of thesis, which is to say late, and for the wrong reason. This is the one task in the operational calendar with no manual number attached to it, because there is no manual version to measure. The cost is counted in positions that outlived their premise. ## The change :: Conditions become monitored items. The change is small at the point of writing and large everywhere after it. Every memo states, explicitly, what would make it wrong. Conditions, not sentiment: the observable facts that, if they occurred, mean the thesis has failed whatever the price is doing. Those conditions then become monitored items on the same lifecycle as every other obligation the firm carries, watched the way a reconciliation is watched. Silent while they hold. Specific when they do not. When a condition trips, it is raised with the original memo language quoted beside what actually changed. The question that reaches the investment team is specific: here is the sentence you wrote, and here is the evidence that it came true. Genba's own reference deployment, a validation estate on private infrastructure running a fictional multi-family-office book, carries a worked instance of it. An allocation memo there reads: "We are backing the founder's own hand on this book; if he stops running it, the allocation has no thesis left." Sharpened into something observable, that becomes two conditions: the named manager stops signing the monthly letter, or a key-person clause is triggered in the fund documents. Seven months pass in silence. Then a monthly letter arrives signed by a newly appointed co-manager, and the item lands in the queue with the memo's own sentence quoted above it. :::embed{component="ConditionWatch"} ::: ## The review :: A tripped condition arrives specific. That framing is the entire value. A general unease about a position produces meetings. A specific tripped condition produces a decision. The team can conclude the condition was drawn too tight and revise it, on the record. It can conclude the thesis has genuinely failed and act. It can conclude the change is real but survivable and write down why. All three outcomes are legitimate, and each of them leaves the firm with a dated, attributable judgment where previously there was silence. The discipline improves the memos themselves. A thesis written in the knowledge that its failure conditions will be monitored gets written more honestly, and vague risks get sharpened into observable ones, because a vague condition cannot trip. ## The machinery :: Why this needed the rest of the system. This task could not be bolted onto a manual process, which is why no firm runs it manually. It needs institutional memory that holds every memo in one reviewed, current canon instead of scattered across drives, which is what [Kyōzō](/insights/kyozo-institutional-canon/) is for. It needs monitoring that runs continuously without consuming anyone's attention on the days nothing changes. It needs an escalation path, so a tripped condition cannot be quietly set aside: the review is mandatory, and it stays open until someone signs a decision. Tōryō supplies both, on the same ladder that chases every other obligation the firm carries. Each of those is general machinery, already carrying the firm's [everyday obligations](/insights/one-queue-obligation-lifecycle/). Pointed at the investment process, together they run the task nobody has ever had the capacity to staff. There is no manual baseline to compare it against, and that absence is the claim. ## References 1. [Kyōzō: the firm should never have to relearn itself](/insights/kyozo-institutional-canon/) — Genba Labs Insights, on the reviewed canon that holds the memos 2. [One queue: the life of an obligation, from inbound to evidence](/insights/one-queue-obligation-lifecycle/) — Genba Labs Insights, on the monitoring and escalation machinery this task borrows 3. [The thirty-day proof: how an investment firm should test agentic AI](/insights/ai-proof-of-concept-for-funds/) — Genba Labs Insights, on testing one workflow against a rubric agreed in advance ## Questions and answers ### How can an investment firm monitor whether its investment theses are still valid? Write the invalidation conditions into the memo as observable facts, then monitor those conditions the way a reconciliation is monitored: silently while they hold, specifically when one trips. A tripped condition is raised with the original memo language quoted beside the evidence of what changed. The review is mandatory, so the position comes up for decision because the reasoning moved, not because the price did. ### Who writes the invalidation conditions for an investment thesis? The investment team, in the memo, as part of the process it already runs. The system monitors what the team wrote and does not invent conditions of its own. Writing a thesis in the knowledge that its failure conditions will be watched tends to sharpen the memo itself, because vague risks cannot trip. ### What if an invalidation condition is subjective? Then it cannot be monitored, and the discipline pushes it toward an observable proxy: a disclosed customer concentration, a filed regulatory decision, a competitor shipping a product, a covenant level. That sharpening is the point of the exercise. A condition nobody can verify was never a usable risk statement. ### Does a tripped invalidation condition force a sale? No. It forces a review. The team can conclude the condition was drawn too tight and revise it on the record, conclude the thesis has genuinely failed and act, or conclude the change is real but survivable and write down why. All three are legitimate outcomes. The fourth outcome, not looking, is the one that stops being available. ### How is thesis monitoring different from a stop-loss? A stop-loss watches the price. This watches the reasoning. The intent is to hear that a thesis has broken before the price says so, and to have the memo's own language on the table when the team decides what to do about it. ### Why do firms not run thesis monitoring manually today? Because it is a continuous task with no deadline, no external chaser and no visible cost when skipped, so it loses to every dated obligation in the calendar. It also needs machinery a manual process cannot supply: one reviewed canon holding every memo, which is what Kyōzō does in Genba's system, monitoring that consumes no attention on the days nothing changes, and an escalation path that makes a tripped condition impossible to quietly ignore.