Ask an investor relations team what distinguishes them from the fund down the road and some version of the same answer comes back. We are responsive. An LP emails on a Friday afternoon and gets a real answer from someone who knows their position, before the weekend, without having to chase.
That is a genuine advantage today. We do not think it will be one for much longer.
The reason is not that IR teams will stop caring about service. It is that the work sitting behind a fast answer is the part compressing fastest. Most LP queries resolve to a lookup and a reconciliation: what is my remaining unfunded, why did the reported NAV move, which fees applied to my vehicle, what did you say about that asset three quarters ago. Answering quickly has always required a person who knew where everything lived and had the bandwidth to go and get it. It is starting to require a system that has already read everything and can show where each number came from.
Speed was never the service, it was the evidence of the service
Allocators do not value the hour for its own sake. They value what a fast and correct answer tells them about the firm behind it: that the records agree with each other, that someone senior is paying attention, that nothing is being quietly reconstructed while the response is assembled. Response time has been a legible signal for a set of qualities an LP cannot otherwise observe from the outside.
Signals work because they are costly to produce. Turning a complex position query around in two hours has meant that a firm invested in people, systems and discipline that a less organised competitor did not. When the cost of producing the signal falls towards zero, the signal stops carrying information. A one-hour response time will soon say about as much as a well-built website says about a manager today, which is nothing beyond the absence of an obvious problem.
This is worth sitting with, because it means the investment most IR teams are currently making is defensive. It buys the right to stay in the conversation. It does not, on its own, distinguish you in it.
The tail collapses first, and the tail is what LPs remember
To put rough numbers on it, and these are illustrative rather than measured: on a mid-market fund with 180 limited partners, an IR function might field somewhere between 800 and 1,200 discrete queries in a year, of which perhaps three-quarters are positional or documentary lookups. Assume a median response of a day and a half. Assume also that the slowest five per cent run to two or three weeks, because they require someone to reconcile a side letter against an allocation schedule, or to establish what a deal team actually meant by a footnote written eighteen months ago.
The median is not where the damage sits. The tail is. An investor who waits three weeks for an answer about their own capital draws a conclusion about your operations, and they are usually right to draw it. Those queries are also disproportionately the ones that matter, because the questions that are hard to answer are hard for structural reasons: the data is inconsistent, the provision is bespoke, the history is in someone's head.
What capable systems do to this distribution is not to shave the median from thirty-six hours to four. It is to make the tail behave like the median. That is the change worth planning for, and it is also the change that removes the last bit of daylight between a well-run IR function and an averagely run one.
A fast wrong answer is worse than a slow right one
The strongest argument against pushing hard on speed is that speed without accuracy is actively destructive, and we think it is correct. An LP who receives an unfunded figure in nine minutes and a correction three days later has been harmed twice: once by the error, and once by learning that your fast answers cannot be relied upon. Restatement is expensive in a way that delay is not. Delay is an operational irritation. A wrong number is a question about whether anything you send can be taken at face value.
Some questions also deserve to be slow. A query about a portfolio company's covenant position, or about why a valuation mark diverges from a comparable transaction, is a question where the considered answer and the immediate answer are different documents. A firm that answers everything instantly will eventually answer something instantly that should have gone to the CFO first.
So the discipline that has to arrive alongside speed is grounding. Every figure traced to the record it came from. Every assertion citable to a source document rather than to a confident summary. And a deliberate decision, made in advance, about which categories of question route through a named person before they leave the building. This is not a brake on the capability; it is what makes the capability usable inside a fiduciary relationship. Firms that skip it will enjoy a year of admiration followed by an uncomfortable operational due diligence meeting.
The next axis is anticipation, and it is less comfortable
If responsiveness normalises, the remaining advantage sits with the question that has not been asked yet. Most LP questions are not random events. They are generated by things the GP already knows about, usually before the investor does.
A material markdown in a named portfolio company, ahead of the letter explaining it.
A capital call landing in a quarter where that investor has already told you their allocation is stretched.
A management fee line that moves because a company left the portfolio mid-period.
A continuation vehicle, a key person change, or anything that will require LPAC consent.
Each of these produces a predictable set of questions from a predictable subset of the investor base. The firm that sends the explanation before the query arrives is not being faster. It is playing a different game, and the LP experiences it differently: not as good service, but as being treated as an insider.
The discomfort is real. Anticipation means volunteering information nobody requested. A great deal of reporting practice is built on the opposite instinct, which is to answer the question asked, accurately, and no further. That instinct has defensible origins in disclosure consistency and in the fear that volunteering to one investor creates an obligation to all. But it is increasingly visible to allocators as a choice rather than a constraint, and choices get judged.
Anticipation is a memory problem before it is a service problem
Knowing what a particular LP will ask requires holding two things in view at once: the fund's current records, and that investor's own history of engaging with them. Which questions they asked in 2022 and again in 2024. Which side letter provisions they negotiated and therefore monitor. Which asset they queried twice. Whether their consultant reads the letter before they do.
Most firms already possess all of this. Almost none of it sits in one place. The parts that reached the CRM are the parts someone had time to type, which systematically excludes the nuanced material and over-represents the meetings that were easy to summarise. The rest is in inboxes, call notes and the recollection of whoever covered the account at the time.
This is the point at which the technology choice starts to carry weight rather than being a procurement detail. Reading a full quarterly pack alongside one investor's multi-year query history, and producing a short list of the three things this particular LP will want explained, is a sustained reasoning task over long and inconsistent documents. It is not a search problem, and tools built for retrieval tend to handle it poorly. It is one of the reasons we build on Anthropic's Claude: the reasoning holds across large document sets, and it can be inspected, so an IR lead can judge whether a prediction is sensible before anyone acts on it.
The firms that get this right will be quieter, not faster
The end state is counter-intuitive. A well-run IR function in five years will not be distinguished by a crowded inbox cleared at speed. It will be distinguished by a thinner inbox, because the explanations arrived before the questions formed. The internal metric changes from response time to something closer to questions avoided, and the external effect is an investor base that is rarely surprised.
That is a better business to run as well as a better one to be invested in. Surprise is what turns a routine quarter into a difficult call, and the absence of surprise compounds quietly into re-up conversations that start from a position of trust rather than reassurance. Responsiveness was always a proxy for that. The proxy is expiring, and what it was standing in for is finally becoming buildable.
