How to Scope an AI Sprint for Board-Pack Prep

It is the Wednesday before the board. Your CFO opens the deck template at 8:12 AM. 42 slides, 11 sections, four function heads with metrics still outstanding, and a Q4 commit page that read 68 percent to coverage at the Monday pipeline call. The CEO wants a draft by Friday 6:00 PM. The investor update goes out Monday 9:00 AM. The board reads the pack Tuesday at 7:00 AM.

By Friday 11:48 PM the deck reads 74 percent complete, three function metrics are still placeholder text, and the CFO has spent 22 hours between the Wednesday open and the Friday close reconciling the same four numbers across Looker, Salesforce, NetSuite, and the Finance Ops spreadsheet nobody wants to admit still runs the actuals. The COO ate 11 hours on narrative. The CEO ate seven on the vision page. 42 hours of senior calendar for one deck that renders on a tablet for 44 minutes.

Every operator asking us how to scope an AI sprint for the pack is asking the right question. The pack is the single quarterly artifact where the stack has to resolve across six functions in one voice, on cadence, with the exceptions pre-cited. It is also the artifact where the queue work is the most visible and the least owned. Here is how we scope it.

What the pack has to hold

A Series B board pack holds four things. The six function metrics against commit. The three cross-function dependencies most likely to miss the quarter. The two decisions the exec team is asking the board to make or endorse. The one risk escalating to the audit or compensation committee.

Every slide in the pack reads against one of those four. The 42-slide version is the full deck. The 18-slide version is the read-ahead. The three-slide version is the executive summary the lead director reads on the drive to the office. The sprint has to produce all three cuts off the same synthesis, so the agent reads the function data once and renders three outputs.

The failure mode we see most often reads the opposite. The CFO owns the pack template, the COO owns the ops section, the CRO owns the revenue section, and the exec team reconciles live on Monday morning. The three cuts never match. The board reads the drift on the second page. The lead director asks the follow-up on the second cycle.

The seven inputs that have to resolve by Thursday

The pack reads against seven systems. Finance in NetSuite or Sage Intacct, revenue in Salesforce, product in Mixpanel or Amplitude, support in Zendesk or Intercom, hiring in Greenhouse or Ashby, marketing in HubSpot or the campaign automation tool, and the Looker or Metabase layer sitting on top of five of the seven. The seven feed the four pack sections. The synthesis is the function.

Finance. The CFO queue reads the Monday close exceptions, the ARR waterfall, and the cash runway against the Q4 plan. The 42 reconciling items from the last close should land on day five of the following close, not Friday night of pack week. The flux brief should ship on day seven with the three largest variances already explained against the prior quarter.

Revenue. The CRO queue reads the 240 open opportunities, the stage-SLA attainment, the forecast reconciliation against the CFO plan, and the three deals most likely to slip the quarter. The forecast and the plan should reconcile on Thursday 8:00 AM, not Friday 11:48 PM. The two expansion deals most likely to close should land on the pack with the champion position already cited.

Operating cadence. The COO queue reads the six function dashboards, the three cross-function dependencies most likely to miss the quarter, and the two decisions the exec team is asking the board to endorse. The 11-hour day is the version where that queue runs by hand. The sprint closes it. The Thursday executive brief rolls up into the quarterly pack with no extra synthesis work.

What day one of the sprint reads against

Day one is scoping, not building. We spend the first 48 hours reading against the last four board packs, the exec-team Slack channel the week before the last three boards, and the comment history on the shared pack template. We land on three things. The output of day one is the sprint spec and the write-access tier.

First, the eleven numbers that have to resolve exactly across all three cuts of the pack. Second, the four narrative paragraphs the CEO rewrites every quarter and the three that have never been rewritten since Series A. Third, the two sections the board has asked a follow-up on in the last three cycles.

Those three findings become the sprint spec. The agent reads against the seven systems, resolves the eleven numbers once, drafts the four CEO paragraphs with the quarter's data pre-cited, and surfaces the two board follow-up sections with the exception already written. The CEO rewrites the four paragraphs in 90 minutes instead of seven hours. The CFO reconciles the eleven numbers in 20 minutes instead of 22. The COO reviews the three cross-function dependencies and signs off.

What ships at day seven and day fourteen

Day seven, the first draft of the read-ahead lands in the CFO inbox at 8:00 AM Monday. The 18 slides render against the seven systems with the eleven numbers resolved, the three function metrics missing from last quarter's draft now cited, and the two slip-risk deals already flagged with the owner and the ask. The CFO reads the draft at 8:09 AM and spends 40 minutes on narrative instead of four hours on reconciliation. The COO reads the operating-cadence section at 8:30 AM and signs off on the three cross-function dependencies before the 9:00 AM standup.

Day fourteen, the Monday brief and the quarterly pack read off the same synthesis. The weekly executive brief lands at 8:00 AM every Monday with the four pack sections pre-cited in miniature. The quarterly pack rolls up from 12 of those briefs with no extra synthesis work. The pack-week calendar drops from 42 senior hours to 11. The Friday 6:00 PM narrative review takes 90 minutes and the Monday 9:00 AM investor update ships on time.

Day one. Spec, scope, access, and the eleven resolve-exactly numbers signed off. Day seven. Read-ahead draft lands in the CFO inbox at 8:00 AM Monday with the slip-risk deals flagged. Day fourteen. Weekly executive brief on cadence, quarterly pack rolls up from the briefs, Friday narrative review takes 90 minutes.

What stays in-house and what the agent owns

The agent owns the queue. The function heads own the judgment. The CFO still writes the audit-committee note on the one risk escalating to the board. The CRO still owns the deal strategy on the three slip-risk accounts. The COO still owns the two decisions the exec team is asking the board to endorse. The queue stops stealing the senior calendar and the function heads land on the architecture work the ledger was paying them to run.

The sprint ships with four access tiers. Read-only on NetSuite, Salesforce, Mixpanel, Zendesk, Greenhouse, HubSpot, and Looker. Write access on the Monday brief draft only. Human approval on every pack-section commit. No agent commits to the final deck without the CFO signoff on the eleven numbers. That split is the reason the CFO signs off on the sprint in week one instead of debating it in week six.

The queue work moves. The judgment stays. The board reads the same voice it has read for three cycles and the lead director stops asking the same follow-up on the second page.

The unit economics on the pack

A $320K CFO reads $420K loaded against the US entity. The pack-week calendar costs the CFO 22 hours, the COO 11, the CEO seven, and four function heads six each, so 64 senior hours a quarter at a loaded hour of $180 to $280. One pack cycle costs the company $11K to $18K in senior time. Four cycles a year costs $44K to $72K and the Friday 11:48 PM slip costs the Monday investor update.

A 14-day sprint to stand up the board-pack synthesis agent runs in the low to mid five figures. Ongoing cost lands at $2K to $5K a month on API spend, read access to the seven systems, and the ops operator who reviews the Monday brief. Year one lands break-even against the queue cost and shifts 48 senior hours a quarter from reconciliation to narrative. Year two drops to the ongoing line and the senior calendar lands back on the architecture work the board hired the exec team to run.

If your Wednesday-before-the-board starts at 8:12 AM and ends at 11:48 PM on Friday with three metrics still placeholder text, the version where the read-ahead lands at 8:00 AM Monday and the Friday narrative review takes 90 minutes is 14 days away. The seven systems still store the data. The 42-slide template still renders on the tablet. The agent closes the synthesis the stack was already supposed to close.

The scope reads simple. 48 hours of day-one access review, seven days to the first read-ahead draft, 14 days to the full cadence. You can keep running pack week on a Friday 11:48 PM commit, or you can scope a sprint and have the next board pack landing before the Monday coffee is cold.