Your morning is not a dashboard. It's a list.
47 KPIs is not triage. Two named actions is.
Across a 30-deal portfolio, every operator opens the morning the same way: a dashboard renders everything that moved overnight, and a person manually decides what matters today. That manual triage is where early warnings get lost. The platform's job is to do the triage upstream — by signal, with named actions, in priority order — before the operator has to ask what to look at first.
The dashboard says everything. The list says what to do.
A monitoring dashboard renders KPIs. It does not tell you which of the thirty positions you hold needs your attention this morning, why, or what the named next action is. That is left to the operator — every morning, by hand, across every deal.
The cost is not visible in the dashboard. It shows up later, as the early warning that nobody surfaced because it sat at line 14 of a 47-row report. The first job of an investment operating system is not visibility. It is to spend its compute budget on the question the operator was about to spend their morning on: given everything that moved last night, what should I look at first, and what is the named action?
A KPI moving is a question. A named next action is an answer. The platform's job is to convert the first into the second before the operator opens the page.
DOORS-A and DOORS-B, side by side.
Both deals are in Portfolio stage. Both look healthy on the monitoring dashboard. Each has a different stale signal — and the platform surfaces the right named action on each, with a priority score, and a single click to the surface that fixes it. 42 seconds, no narration.
What the platform did before the operator opened the page.
- 01 · 00:00Two deals, both in Portfolio.
DOORS-A and DOORS-B. Same stage, same fund, same operator. Nothing on the surface tells you which one needs attention first.
- 02 · 00:08DOORS-A: 90-day-stale actuals.
The lifecycle probe sees the actuals snapshot is 90 days old against a 45-day threshold. The next-action engine surfaces a named action — Upload actuals — at urgency 0.75. One click to the actuals upload form.
- 03 · 00:25DOORS-B: covenant test 60 days past cadence.
Actuals are fresh; the Min DSCR test was last run 60 days past its 30-day cadence. The engine surfaces Refresh covenant test — Min DSCR at urgency 0.72. One click to the covenant runner.
- 04 · 00:38Both signals carry a priority score.
The two actions sit on a single ranked list — stale actuals (0.75) ahead of stale covenant (0.72) ahead of cycle-review noise (0.65). The operator did not have to look across thirty deals; the platform already did.
The triage is deterministic, not opinion.
The next-action engine scores every candidate signal against a fixed urgency model and returns the ranked list verbatim, with the inputs that fed the decision. Nothing is hidden behind a black-box ranking. The operator can audit why one signal beat another and override it where their judgment differs.
- ▸Lifecycle probe — every Portfolio-stage deal evaluated against actuals freshness, covenant cadence, IC condition status, cycle review timing, and ten more signals
- ▸Urgency-scored ranking — each candidate carries a fixed urgency score (0.0 to 1.0). Ranking is deterministic and auditable, not learned
- ▸Named action — every signal is bound to a specific next step (Upload actuals, Refresh covenant test, Clear IC condition) and a route the operator can act on directly
- ▸Decision rationale — the top action is rendered alongside the runner-up alternatives, with the reason each lost. The operator sees what was suppressed and why
- ▸Source-document lineage — every figure that fed the decision traces back to its source page; the morning list is auditable from the headline action down to the original document
- ▸IC anchor + since-IC drift — the morning list surfaces what has changed since the last decision-commit anchor, not just what is true today
The video is two deals. The real thing is your portfolio.
Send a position from your portfolio and we'll walk it through the loop together — 60 to 90 minutes, on your real data. You'll leave with the morning list rendered against your own deals.