AI automation for fund operations and investor reporting in the UAE.

The NAV pack is built on sources that already agree, and the number is settled before it is signed.

The administrator’s file, the custodian statement, the bank records and the trade file are read as they arrive and matched against each other, so the pack is built on records that already agree.

What disagrees is named with both sides shown, and anything that moves the number stops at the person who signs it.

Book a meeting

Thirty minutes. Nothing to prepare.

Or email hello@runbook.ae

The work

Four runbooks for a fund.

The NAV pack
Struck on sources that agree, not on sources that are chased.
Prices, positions, cash balances, accruals, the fee calculation. They arrive from the administrator, the custodian and the bank, each on its own schedule and in its own shape. They are read as they land and matched against each other. Anything that disagrees goes to the person who signs the NAV, with both sides shown.
The investor report
Out on the date, in the shape each mandate asked for.
Performance, exposure, capital account balances, the commentary. They come from different places, and every mandate wants them presented differently. The assembly runs from the same agreed figures every period. What reaches an investor is what the fund actually signed.
Capital calls and distributions
Every notice ties back to the commitment it draws on.
A call or a distribution means allocating across the register and checking each amount against the commitment and the terms behind it. The right notice, with the right instructions, has to reach the right investor. The notices are assembled and matched back against what actually arrives. A shortfall is raised while it is still this period’s business.
The audit and administrator requests
Answered with the workings already attached.
The auditor’s request list and the administrator’s queries land against a period that has already closed. Each answer means finding the document, the calculation and the approval behind one line. The support is assembled as the work happens. A request is answered from a file that already holds the workings.
The alternative
Fund systems end at the record.
They hold the positions and the balances, and the work of making the sources agree still happens in someone’s head. Runbook arrives knowing your business, because writing down how your NAV is actually struck is where we start.
What you own
You own it. No seat bill.
It runs in your cloud account. If you stop paying us, it keeps running. What stops is us watching it, fixing it and extending it. The model and the data location are chosen by you and written into the build.
What it has to pass
Nothing goes live until it passes your own hardest cases.
We build the test set from the NAV packs and reports your team argued about, freeze it before the build, and run it again after every change. It is yours, and another team can be measured on it.

What we get asked

Questions a fund COO asks.

A great deal sits on the fund’s side of that relationship. What gets sent to the administrator, what comes back, and the check between the two before a NAV is agreed all stay with your team, along with the investor reporting, the calls and the distributions built on top of the number. Runbook builds and runs the agents that do the fund’s side of that work, and your administrator carries on doing theirs.
No. Every statement, notice and report stops at a named person for release, and the system’s job is to have it complete, checked and traceable by the time it gets there. The line is drawn with you during the build, and anything the agents cannot resolve stops there with the reason attached.
That is the normal starting position and it is where the work begins. The data layer is built first, so positions, cash, fees and investor records are read where they already sit, including the spreadsheet one person maintains and the report that only ever comes out as a PDF. Nothing has to be replaced for the first workflow to run.
Those are the reason this is encoded for you rather than bought off a shelf. The rules are written down as they actually work, including the exceptions nobody has documented, and the system runs them exactly. Where the rules run out, it stops and a named person decides.
The scope, the fee and the finish date are agreed with you in writing before anything starts. We do not publish a length, because it follows the size of the operation, how many systems the work touches and what condition the data is in.
You do. The code, the credentials and the documentation sit in your accounts and under your name from the day we start. It runs in your cloud account, and if you stop paying us it keeps running.
Nothing. It is thirty minutes with the people who would build it: the job as it runs today, where the line would sit, and what we would measure.
It runs in your own cloud account, in the region you choose, under your own credentials. Every source the system is allowed to reach is declared during the build, and anything not declared is not reachable when it runs. That declaration is a document you can hand to an auditor or a regulator.