At Alva Labs the monthly board report used to take days. It now assembles in seconds, because an agent with read access to the ledger and the CRM builds it on request.
The interesting part is not the speed. It is what the days were being spent on. Almost none of that time was analysis. It was retrieval: pulling numbers out of 3 or 4 systems, reconciling the versions that disagreed, and rebuilding the same pack structure that was rebuilt the month before.
Management reporting is slow for a structural reason, and no amount of dashboard software fixes it, because the dashboard is downstream of the problem.
A board report that took days at Alva Labs assembles in seconds, across 3 countries on 1 ledger.
to assemble the monthly board report at Alva Labs, built by an agent on request
of Light's discovery calls that raise reporting and visibility as a problem
entities at All Gravy reporting on 1 chart of accounts, with department and country as dimensions
Two numbers, one company
The recurring failure looks the same everywhere. There is the number the board sees, built in a warehouse or a spreadsheet from CRM and product data, and there is the number in the ledger, and they do not agree.
Neither is wrong. They are built from different sources on different cadences by different people, so they answer slightly different questions and diverge by a few percent. That divergence costs a day a month to reconcile and a great deal of credibility the first time somebody notices it in a board meeting.
Dreamdata's Peter Egehoved has a name for the condition: two world syndrome, which surfaces every time an auditor asks him to walk from an ARR number to audited revenue. The reconciliation is only necessary because the 2 numbers were derived independently in the first place.
Reporting is a property of the chart of accounts
The unglamorous truth is that most reporting problems are structural problems in the ledger, showing up at the reporting layer where they are most visible and least fixable.
All Gravy is a clean example. Its old chart of accounts was built for a company it no longer was, years before it became a SaaS business, and department reporting ran on dedicated GL accounts rather than dimensions. Marketing spend needed its own account rather than a tag. Any report cutting across department and country had to be assembled by hand, because the structure could not express the cut.
During migration the team rebuilt the chart of accounts from the ground up for a SaaS business, with custom dimensions for department and country in place from day 1. That single decision does more for reporting than any BI tool bolted on top, because it makes the cut the business actually wants a query rather than a project.
Old model
Assemble, reconcile, present
Pull from several systems, reconcile the versions that disagree, rebuild the pack. Most of the elapsed time is retrieval, not analysis.
Light's model
Read the ledger directly
Entity, department and country sit on the transaction. An agent with read access builds the pack on request from records that already agree.
What changes
Analysis while it matters
The pack lands early enough in the month for the numbers to change a decision rather than describe one already taken.
Timeliness is the whole point
A report that is accurate and late is a historical document. The only reason to produce management information is to change a decision, and a decision made on day 20 cannot be improved by a pack that arrives on day 21.
This is why continuous close and management reporting are the same subject. Tillo went from a 12 day close to 5, All Gravy closes 4 countries in 5 days, and Oper Credits stopped consolidating at month-end entirely because it posts as transactions happen. Each of those is a reporting improvement before it is an accounting one, because each moves the moment the business can see itself closer to the moment things actually happened.
Astra, the always-on analyst, sits on the same principle from the other direction, watching what agents and humans post across the ledger and raising what needs attention rather than waiting for someone to ask.
What good looks like
Three tests, and they are harder than they sound.
Can somebody produce a group view across every entity without assembling it? At All Gravy that means 4 countries in 1 place, which its finance analyst named as the single change that most altered her day.
Does the ARR number in the board pack trace to a contract, an invoice and a payment in the same system? If it requires a bridge, the company is carrying 2 stories about its own revenue and paying for the reconciliation monthly.
Can the pack be regenerated on request rather than rebuilt? Alva Labs' answer is seconds. A team that cannot regenerate its reporting is maintaining a document, not a reporting function.
Reporting stopped being the last task of the month at these companies. It became a view of a ledger that was already current, which is a different thing entirely from a faster way to build a pack.
See how reporting works in Light, or book a demo.