Investor Relations
Our Services
We identify where AI removes cost and where it quietly adds risk.
Every engagement begins with a scoping conversation. If a smaller package is the right starting point, we will say so.
Your LPs ask harder questions, more often, and in formats that no longer standardise. IR teams absorb the difference by hand: the same capital account query answered differently depending on who picks it up, the quarterly letter assembled from scattered drafts, the portal request that lands in the middle of a close. We design AI-assisted IR workflows across ad-hoc requests, quarterly letters, portal responses and capital account queries, built on Anthropic's Claude and wired to the systems of record you already run. Response times improve and answers stop diverging. What does not change is the review step: a named person signs off, and no unreviewed number reaches an investor.
Answering, Drafting and Checking
Most IR requests are retrieval problems wearing a drafting problem's clothes. We build workflows that route an incoming query to the sources that can answer it, the capital account statement, the partnership agreement, the side letter register, the prior response to the same question, and produce a draft that cites each figure back to where it came from. Where two sources disagree, the draft says so rather than choosing. Quarterly letters are assembled the same way, from portfolio commentary, the administrator's performance file and last quarter's approved language.
The line of judgement sits at the number. The system drafts, retrieves and cites. It does not assert a figure it cannot trace, and it does not paraphrase a contractual obligation. Anything touching valuation, forward-looking commentary, or a figure that differs from the administrator's record goes to a named reviewer before it reaches an investor. The reviewer's time then goes on judgement rather than on finding the answer again.
ad-hoc lp requests
quarterly letter assembly
capital account queries
portal response drafting
investor onboarding checks
How We Scope IR Work
We start with a Diagnostic: a paid, time-boxed review of what the IR function actually handles across a reporting cycle. We read the request log, the last set of letters, the portal traffic and the capital account queries, and return a written map of which request types recur, which are supported by data you can stand behind, and which should stay manual.
The Pilot takes one of those types, usually the highest volume and the lowest judgement, and runs it narrowly with a named owner and a measure agreed before it starts. Build puts the surviving workflow into production against your CRM, administrator feed and document store, with the review checkpoint designed in rather than added afterwards. Run covers monitoring, the governance record your compliance function and your investors' operational diligence teams will ask for, and periodic re-scoping, because what your LPs ask changes with each fund.
Diagnostic: map a full reporting cycle of requests, letters and portal traffic
Pilot: the highest volume, lowest judgement request type, with a named owner
Build: production against your CRM, administrator feed and document store
Run: monitoring, the governance record, and re-scoping as LP questions change
