When we started Jolv, one question kept coming back: why does so much valuation work have to wait until quarter-end?
A portfolio does not move quarterly. A company sends its accounts in March. A listed peer reports in May. A competitor is acquired in June, at a multiple everyone notices. A CEO answers the questionnaire the day it reaches her. None of this waits for the 30th of June. The valuation does.
Not because it should. Because the team has never had the capacity, or the tooling, to process the work when the information arrives. So it piles up, and the quarter closes in a rush. I wrote about where that bottleneck actually sits in an earlier note: rarely on method, almost always on scattered inputs. This note is about the other half of the answer.
What "continuous" means, and what it does not
Let me start with what it does not mean, because the word invites misreading.
Continuous valuation does not mean a fair value that drifts every day with the market, with nobody deciding anything. It does not mean an AI that values a company on its own. A valuation remains an opinion, formed by people, at a date, under a written policy, and defended in front of a committee and an auditor. That does not change, and nothing in our work is meant to change it.
What changes is when the work gets done. In a continuous process, every building block of a valuation moves forward as soon as its input is available. The accounts land in the campaign the day they are received. The comparable set refreshes when a peer reports. The monitoring flags the acquisition the week it happens, not the week the analyst remembers it. By the time the valuation date comes, the file is already assembled, reviewed piece by piece, and the remaining work is the one that deserves a human: the judgment call, the review, the approval.
The quarter still exists. It becomes a date on which a value is signed off, not a month during which everything is rebuilt.
Three conditions
We have learnt that this only works if three things hold at the same time.
The inputs come in without being retyped. Management data arrives through a secure link and lands directly in the campaign. Listed comparables come from a market data feed. Transaction multiples come from a shared base. The monitoring reads the press and the legal registers on each portfolio company and classifies what matters. If any of these still goes through a spreadsheet pasted by hand, the process is not continuous, it is merely earlier.
Every step is attributable. Who retained which multiple, when, replacing which previous value, with what justification. This is what the IPEV guidelines expect when they ask a valuer to be able to reconstruct what was known and knowable at each valuation date, and it is what backtesting relies on. In Jolv, every change is versioned, time-stamped and attributed. A correction becomes a new version, never a rewrite.
The approved result is frozen. Once a campaign is reviewed and approved, it does not move. We call it Freeze & Lock: the numbers presented to the committee are the numbers in the file, six months later, when the auditor opens it. Without this, a continuous process is a liability, because nothing distinguishes the value that was signed from the value that kept moving.
The first condition is about capacity. The second and third are about trust. A continuous process that is not attributable and not frozen is faster, and worse.
Where people step in
This is the part I care most about, because it is where the industry conversation on AI tends to go wrong.
Agents bring capacity. They move files forward, integrate data, refresh sets, flag changes, draft the first version of a memo. They do not sign. The judgment on a retained multiple, on where a liquidity discount applies, on whether a provision is warranted, on whether a cap table waterfall is being read correctly, stays with the people whose name is on the valuation policy. The workflow is built around that: a valuation is submitted, reviewed, approved, then frozen, with separate roles at each step.
The point of removing the month-end rush is not to remove people from the process. It is to give them back the time the rush was taking, so that it goes into reviewing, understanding and deciding.
Where we are
I would rather be precise than impressive here.
What Jolv does today: valuation campaigns by period, management data collected through a secure link and integrated into the campaign, listed comparables and transaction multiples, the equity bridge from enterprise value to the fund's share, DLOM applied where the IPEV 2025 guidelines place it, the submission, review and approval workflow, the immutable audit trail, Freeze & Lock, and daily monitoring on portfolio companies. Everything described above as "arriving" or "refreshing" relies on these.
What we are building, with the funds we have been talking to since the start of the year: the remaining pieces that let a campaign stay open between two valuation dates and absorb information as it comes, rather than being created at quarter-end. They pushed us to solve the hard parts properly, and they shaped much of what Jolv has become. We have no paying customer to name yet, and I will not name one until there is.
The broader picture, the architecture and the standards we build against, is on the product vision page. This note is the shorter version: the reason behind it.
Your portfolio moves. Valuation can now move with it.