The board pack as a derived artifact.
The board pack is the most scrutinized document a SaaS CFO produces. It is also, in most companies, the most manual. The analyst spends a week assembling it from sources that disagree by tens of thousands of dollars. The CFO spends a Sunday reading it for the first time. The board reads it for thirty minutes and asks a question the spreadsheet cannot answer.
HarborOS treats the board pack as output, not artifact. The number on the cover slide is the number Compass locked on Friday. The narrative writes itself from the snapshot diff. The cohort retention table the directors keep asking for is a one-click view in Compass — not a two-week build. And the audit trail PE has always wanted is the snapshot itself.
It is the difference between a deck the CFO carries into the room and a deck the CFO can defend in the room.
The board pack is a manual artifact built under deadline pressure.
The week before a board meeting in a PE-backed software company tends to look the same in every office. The analyst is rebuilding the ARR bridge from Salesforce, NetSuite, the renewals tracker, and the customer health export. The VP of Finance is reconciling three numbers that disagree by a hundred thousand dollars and writing a footnote that says they reconcile. The CFO is writing the narrative paragraph from scratch on Sunday evening, working off a printout of last quarter's deck and a stack of post-it notes.
Monday morning, the deck goes out. The board reads it. A director who has been doing this for twenty years asks for net dollar retention by ARR band, sliced by acquisition channel. The view does not exist. Someone says "we can get back to you on that." Two weeks of analyst time later, the view exists in a one-off file that will never be opened again.
The variance commentary in the deck reads "ARR grew $4.2M this quarter, driven by new logos and modest expansion, offset partially by a single large contraction." It is true. It is also unspecific. The CFO knows which contracts the sentence is referring to. The board does not. PE notices. The next board meeting begins with a request for "contract-level commentary going forward" and the cycle starts again.
Three months later, the board asks the question every board asks: what did we believe in February? The February deck is a PDF. The model behind it has been edited fifteen times since. There is no way to recover the answer except to ask the analyst — and the analyst no longer remembers.
The board pack reads off Friday's snapshot.
HarborOS replaces the assembly with a read. The board pack is rendered from the same locked snapshot the operating team works against every Monday. The narrative is auto-generated from the snapshot diff. The cohort and slice views the directors ask for are filters on the same dataset. The audit trail is the snapshot itself — immutable, dated, signed.
The board pack reads off the same snapshot the team locked on Friday. There is no separate model. No re-aggregation. No "board version" of ARR that disagrees with the operating version. The number on the cover slide is the number in Compass. The bridge in the deck is the bridge in Compass. The deck is a render of the system of record — not a parallel one.
The narrative writes itself, then the CFO edits. Every material variance is traced to specific contracts. The auto-generated paragraph reads "ARR increased $211K week-over-week, driven by Expice B.V. ($166K, new logo) and DataForge ($28K, uplift)." The CFO edits the tone, adds context, removes what the board does not need. The numbers do not need checking — they were the source.
Cohort views and slices are not a custom build. Every snapshot can be filtered by cohort, by entity, by ARR band, by acquisition channel, by signing quarter. When the director asks for net dollar retention by ARR band sliced by acquisition channel, the view is rendered in Compass on the spot. "We can get back to you" becomes "let me share my screen."
The audit trail is the artifact.Every snapshot is dated, signed, and immutable. When the board asks "what did we believe in February?" the February snapshot is exactly what was reported in February — not a reconstruction, not a best guess, not a forensic exercise. Beacon exports it on demand. PE has the audit trail it has always asked for. The CFO has the defense.
What changes when the board pack is rendered, not assembled.
| Before HarborOSAssembled under deadline. | With HarborOSRendered from a snapshot. |
|---|---|
| Board pack assembled manually the night before. The CFO reads it for the first time on Sunday. | Board pack reads off Friday's locked snapshot. The CFO reviewed the underlying numbers on Friday morning. |
| Narrative written from scratch the night before. The variance commentary is generic. | Narrative auto-generated from the snapshot diff, then edited by the CFO. Every dollar of variance is named. |
| Cohort and slice views are one-off Excel builds. The director's question takes two weeks to answer. | Every snapshot is filterable by cohort, entity, and band. The director's question is a one-click view. |
| "What did we believe in February?" requires a forensic exercise. The model has been edited fifteen times since. | February's snapshot is the answer — locked, dated, immutable. Exported on demand. |
| Variance commentary doesn't reconcile to specific contracts. The footnote is "ask the analyst." | Every variance lane is a list of contracts. The footnote is the data. |
| New director asks for a view; the analyst builds it for two weeks. The view is never reused. | New director asks for a view; Compass renders it. The view is reusable for the next board. |
What the board actually gets.
The board pack is six artifacts. Each one is derived from the same locked snapshot. Each one is rendered, not assembled. Each one is exported by Beacon on a schedule, and each one carries the signature of the snapshot it was rendered from.
ARR Bridge
The waterfall the board reads first: starting ARR, plus new, plus expansion, less contraction, less churn, plus or minus FX, ending ARR. One slide, dollar-exact, reconciled to the contract base by construction. Click a lane in the live deck and see the contracts.
FormatSlide · PDF · CSVLocked Snapshot
The reforecast as of board date. Immutable. Comparable to the prior snapshot, the budget, and the rolling forecast. Carries the date, time, and signature of the CFO who locked it. The board pack is rendered from this object.
FormatSnapshot · PDFVariance Narrative
A short paragraph explaining every material move since the prior snapshot. Sourced from the contract diff. Names the customers. Quantifies the dollars. Edited for tone by the CFO. Survives scrutiny because the underlying contracts can be opened.
FormatParagraph · SlideCohort GRR & NDR
Retention by signing quarter. Net dollar retention by ARR band. The PE-pushed views, rendered on demand. The cohort table is not a custom build — it is a filter on the same dataset that produced the bridge.
FormatTable · HeatmapRisk Panel
Renewals at risk in the coming quarter, confidence-weighted, tied to the contract review record. The board sees the exposure and the work plan. Not "we are watching it" — the customers, the dollars, the timeline.
FormatSlide · ListAudit Trail
Every contract change since the last locked snapshot. Who, what, when, why. The trail PE has always wanted, available as an appendix or a separate document. Sent on request — never reconstructed.
FormatLog · CSV · PDFA board pack you can defend.
One source. One snapshot. One number.
The narrative is in the diff.
The trail is in the snapshot.
Locked Friday. Sent Monday. Defended in the room.