Every quarter, a group of capable people spends two or three weeks assembling a document that most of its readers will skim once and file. The quarterly letter survives not because allocators find it the most useful thing a general partner produces, but because it is the thing general partners have always produced.
That is no longer a neutral fact about industry convention. Allocators have quietly stopped treating the letter as their primary means of understanding a position. They read it, they file it, and then they ask the questions they actually had, by email, on a call, or through a bilateral data request. The letter has become the ceremony. The real reporting happens elsewhere, ad hoc, unmeasured and largely invisible to the GP's own operating metrics.
The honest way to describe the quarterly cadence is that it was a constraint of the medium dressed up as a discipline of the reader. It is worth asking what happens to that convention now that the constraint has gone.
The cadence was set by the cost of assembly, not by the needs of the reader
Consider what producing a letter has historically involved. Someone pulls fund-level numbers from the administrator. Someone else chases portfolio company commentary from deal teams who are mid-process on something more urgent. Valuations arrive late from the valuation file. Attribution is rebuilt in a spreadsheet. Three partners mark up the draft, each in a different register, and a fourth version reconciles them. Compliance reads it. Design lays it out. Somewhere in there, a member of the investor relations team notices that a portfolio company's revenue figure does not tie to the number in the last letter, and spends a day working out which one was wrong.
Quarterly was simply the shortest interval at which an organisation could sustain that without the exercise consuming the entire function. Monthly would have broken people. Annual would have looked negligent. Ninety days was the equilibrium between production cost and perceived diligence, and it hardened into a standard.
What it was never calibrated to was the allocator's own clock. An investment committee at a pension plan does not meet on a schedule that aligns with your letter. Pacing models get updated when the team gets to them. Cash forecasting runs monthly at best. Re-up decisions are made when they are made. Board reporting at the LP has its own deadlines. Almost nothing in an allocator's process runs on a ninety-day drumbeat that happens to match yours. The cadence was the GP's production constraint, exported to the LP and relabelled as a service standard.
Assembly is now the cheap part of the job
Nearly everything in a quarterly letter already exists somewhere inside the firm before anyone starts writing. Portfolio companies report monthly. Board packs are circulated. Deal teams update memos. The administrator holds capital account data. The valuation file exists. The previous four letters exist. The work of producing a letter has never been invention; it has been collection, reconciliation, and rewriting a dozen inconsistent sources into a single voice.
That is precisely the category of work that capable language models now do well, and it is worth being specific about why. The useful capability is not fast text generation, which was never the bottleneck. It is sustained reading across long, badly structured, mutually inconsistent document sets: holding a three-hundred-page board pack, four prior letters and an administrator's statement in view at the same time and noticing that a portfolio company changed its revenue recognition definition in the second quarter and that the change was never reflected in the commentary. That is the work that used to take a senior associate a week and that nobody enjoyed. We build on Anthropic's Claude models specifically because that cross-document consistency work is where the quality difference shows up, and where a weaker system produces confident nonsense that an IR team then has to catch.
Once assembly is cheap, cadence stops being a constraint and becomes a choice. And a choice, unlike a constraint, has to be defended.
Valuation genuinely is quarterly, and pretending otherwise would be a mistake
The strongest argument against all of this deserves to be made properly, because it is a good one. Marks move on a quarterly cycle for sound reasons. Valuation committees meet. Auditors expect methodological consistency. Comparable company data arrives with a lag. Independent valuation providers work to a calendar. A general partner who published a fresh net asset value every fortnight would not be more transparent; they would be manufacturing precision that the underlying methodology cannot support, and inviting allocators to model noise as signal.
The same holds for judgement. A partner's considered view of a portfolio should not change weekly. If it does, that is information of a very different kind. The reflective quality of a good letter, the part where someone says what they got wrong and what they now believe, benefits from the space between reporting dates. Allocators are not asking for twelve letters a year, and their own teams are already stretched.
So the argument is not that GPs should report more often. It is narrower and, we think, harder to dismiss: valuation cadence and information cadence were fused because a single expensive process produced both, and there is no longer any reason for them to be fused.
Continuous does not mean constant
Separating the two streams is the whole of the practical change. One stream is positional and factual. It concerns what is true about an LP's relationship with the fund at a given moment, and it does not require a valuation committee to be true.
Commitment, drawn capital, remaining unfunded, and any recallable amounts
Distribution history, split between return of capital, gain and recallable
Management fee and expense detail, with any side letter offset applied
Notice history, including what was sent, when, and whether it was acknowledged
Current portfolio holdings at last published mark
None of that needs to wait ninety days. Today it usually does, or it arrives because someone emailed the IR team and a person assembled it by hand.
The other stream is judgement: what we believe is happening, why we acted as we did, what concerns us, what we expect. That belongs on a slower clock, and arguably on a slower clock than it currently occupies.
The asymmetry between two GPs is what will force the change
Conventions rarely change because someone argues they should. They change when the comparison becomes uncomfortable. To put an illustrative shape on this, and these figures are modelled rather than measured, an allocator with eighty active manager relationships receives in the region of three hundred and twenty letters a year, alongside capital account statements, notices and consent requests. Their capacity to read carefully is finite and they allocate it to whatever is easiest to interrogate.
Now place two GPs in the same strategy side by side. An LP asks both the same positional question in the second week of a quarter. One answers the same afternoon from live records, with the figures traceable to source. The other explains that the information will be included in the next quarterly report. The second GP is not perceived as disciplined. They are perceived as disorganised, or as guarded, and neither impression is corrected by pointing out that quarterly reporting is the industry standard. Operational due diligence teams have already started probing data availability directly, which is the same question asked in a more formal register.
That perception is not entirely fair. It is, however, the one that forms.
The letter gets better once it stops carrying the data
The constructive version of this argument is not that the quarterly letter disappears. It is that the letter is finally relieved of a job it was never well suited to.
A letter that no longer has to transmit positional data can stop being a numbers delivery vehicle with commentary wrapped around it. It can become what the best letters already are and what most are prevented from being by the mechanics of their own production: a genuine argument about the portfolio. What we bought and why we believed it. What is not working, what we are doing about it, and what would make us change our minds. Where last year's view proved wrong. Shorter, more candid, more clearly written by someone with a view rather than assembled by a committee under deadline.
The firms that get there first will have done something fairly simple. They will have decided, deliberately, what information should be continuously available and what deserves a periodic argument, and they will be able to explain the reasoning to an allocator who asks. That is a better position to be in than defending a ninety-day cycle on the grounds that everyone else has one, particularly when everyone else is starting to notice that the reason for it went away.
