All work

LP Reporting Workflow

Private Capital / Investor Relations · Agentic AI

A workflow that pulls performance data straight from the systems holding it, normalises it into one shape, flags the figures that disagree, and has an agent draft the LP report from the confirmed numbers, ready for review.

[ fig. 1 ]A walkthrough of the LP Reporting workflow: a fund's files uploaded at once, conflicting figures flagged for review rather than guessed, and the quarterly report drafted. Recording, 0:06–0:49.

16 of 21

Planted traps caught in the test fund documents after the 18 May tuning, among them a 7.7% headcount gap between two files and ARR that does not match annualised revenue.

A test score on sample documents, not a client’s figures.

The problem

Quarterly LP reporting took three weeks. Fund performance data was pulled by hand out of the systems it lived in, reconciled in spreadsheets, and the report written from scratch every cycle — while ad-hoc investor questions waited days for an answer.

Before
Three weeks
After
Scoped to three hours

Three weeks to three hours is the target the build was scoped to, not yet a measured cycle.

How it’s built

05 · REVIEWYour team reviews itThe draft arrives ready for review04 · DRAFTAn agent drafts the LP reportWritten from the confirmed numbers03 · FLAGFlags the figures that disagreeSent to review rather than guessed02 · NORMALISENormalises it into one shapeEvery source, one set of figures01 · SOURCESReads the fund's own systemseFront, Allvue, FundCount, NetSuite and more
[ fig. 2 ]A drawing of the system, not a screen. Swipe sideways to see all of it.

In numbers

25
LP report endpoints
27
KPIs extracted
2
report formats

Stack

Built with

  • Next.js
  • TypeScript
  • Python
  • FastAPI
  • PostgreSQL
  • Qdrant
  • Redis
  • Azure
  • MCP

Connected to

  • eFront
  • Allvue
  • FundCount
  • NetSuite
  • Sage Intacct
NextIssuelyLegal / Banking & Finance