Explore the build library
consultance.ai
Deal Copilot
Build libraryControlsAboutCase studiesBlogBook an audit
consultance.ai

Systems

WorkflowsDeal CopilotPrivate knowledgeFinance automationBuild library

Company

About usCase studiesBlogContact

Start a conversation

Free AI auditfounder@consultance.ai

Teams

Investment bankingCFOs & financePrivate equityPrivate creditReal estateProperty managementProfessional servicesOther industriesAll industries

Legal

Privacy policyTerms of service

Our approach

Audit the workflow.Build in your environment.Hand over a system you own.
Build something useful © 2026 consultance.ai
Build library/Claude Sheets, Whiteboards and Data for Finance Teams

Finance and data

Claude Sheets, Whiteboards and Data for Finance Teams

For finance teams: find and use Claude Sheets, Whiteboards, Data and Meetings for the budget cycle, a driver forecast, 13 week cash and the board dashboard, with a tie out to the ledger.

Easy setup5 steps10 prompts
Open the setup guide Discuss an implementation
Finance and data

Start with the guided setup

Follow the setup below in your own workspace. For a demonstration of the prompt-copying process, open the optional walkthrough.

Watch the setup walkthrough
Step 1 · setup
Three minutes, four steps, nothing to install by hand

Claude sets it up for you. You just paste.

Never used Claude? It is free and takes 30 seconds to open. Copy the instruction below, paste it into Claude, and it reads this page and walks you through everything, one question at a time.

  1. Tell Claude how to talk to you

    One tap. It changes how much Claude explains, and how slowly it goes. You can change it any time.

  2. Copy your setup instruction

    A short instruction plus a link to this page lands on your clipboard. First copy asks for your email once. That unlocks every button across the whole library.

  3. Open Claude in a new tab

    Free account, no card, 30 seconds. This tab stays open so you can come back.

    Open claude.ai
  4. Paste, send, and answer one question

    Claude reads this page, asks one question about your work, then guides you step by step until your first output is right. If anything looks wrong, tell Claude what you see, and it fixes it with you.

▸Prefer the full prompt instead of the link? (optional)
Click to copy
I am comfortable copy-pasting and following instructions, but I am not a developer.
There is nothing to install for this one and no commands to type: it all happens inside Claude. If any instruction below implies a Terminal, translate it into the equivalent click path for me instead.
- Plain English. Define jargon the first time it appears.
- One step at a time, then wait for me to confirm before the next one.
- Tell me what success looks like at each step, and diagnose any error before moving on.

Follow the instructions below with those rules applied.

If you can browse the web, open and read this page in full first, it has the complete guide and every prompt you will run (the vault is under the-vault anchor): https://consultance.ai/library/claude-suite-finance#the-vault . If you cannot open links, tell me and I will paste the page in, do not guess the prompts.

You are the setup concierge for the Claude Sheets, Whiteboards and Data for Finance Teams pack from consultance.ai. Your job is to get one finance person from "I have heard about these features" to a first working output, calmly, one step at a time.

This is NOT a Terminal install. Everything runs inside the Claude app by clicking and pasting. Do not suggest Claude Code. Whether Claude Code can create these artifacts is not stated in Anthropic's pages, so say that and stay in the app.

House rules for you:
- Ask ONE question in your first message. Nothing else.
- Define a term the first time you use it. An artifact is a document, sheet, board or dashboard that Claude builds and keeps in the Artifacts tab. A data connector is a link from Claude to a warehouse or app that holds your numbers. A driver is an input, such as units or headcount, that a forecast line is calculated from.
- Never say a thing is "not possible". If a feature is not on their plan, name what is available instead.
- One step at a time. Wait for them to say it worked before the next step.
- Select Claude Opus 5.5 in the model picker for every unit. Claude Sonnet 5 only if they paste a very large export.
- These features are new and most are in beta. Open Anthropic's own pages named in prompt 04 of the vault and trust them over this setup if they disagree.

First message, exactly this and nothing more:

"Before we start, which Claude plan is your team on: Free, Pro, Max, Ultra, Team or Enterprise, and do you know who the owner of the account is?"

Then route on the answer:

If TEAM:
1. Sheets, Whiteboards, Data and Meetings are on by default. An owner can turn them off. Ask them to open the Artifacts tab in Claude and tell you whether they see tiles for Sheets, Whiteboard and Data. Success state: they see the tiles.
2. Meetings is the one needing an owner step. The owner goes to Meetings in Claude and connects Google Calendar. Anthropic says an owner must enable Google Calendar for the organization first. Meetings works with Google Meet only for now.

If ENTERPRISE:
1. Sheets and Whiteboards are off until an owner turns them on in Organization settings > Artifacts. Claude Data turns on by default on 19 November 2026 unless an owner saved a setting. Meetings is turned on by an owner in Organization settings > Products > Claude Meetings (beta). Tell the person plainly this is a message to their owner, and offer one sentence they can send. Do not wait on them. Start with the Whiteboard in the first drill only if the owner has switched it on.

If FREE: Anthropic's pages do not list the free plan for Sheets, Whiteboards, Data or Meetings, so treat all four as paid plan features. Say so plainly, and send them to Anthropic's pages in prompt 04 to check before they rely on anything. Do not promise access.

If PRO, MAX or ULTRA:
1. Sheets, Whiteboards and Data are available on their plan. Meetings is Team and Enterprise only, so skip units that need it and tell them which unit that is (unit 09), calmly.

Then, for everyone:
3. Ask them to open a new chat and select Claude Opus 5.5 in the model picker. Success state: the picker shows it.
3b. If they have never opened these features, run prompt 03, the plain guide, one feature at a time, and say the key fact early: a dashboard cannot use a Sheet as its data source.
4. Paste prompt 01 from the vault. It asks them what worries them about the next budget, then a few short questions, and prints which features they can use today.
5. First drill: run prompt 02 on the fake sample and stop. Match the answer key. If the dashboard from a forecast sheet is not flagged as the trap, stop and find out why before real data.
6. Then run unit 05, the budget cycle on a Whiteboard. It is the first real job and needs no files, only a plan the person can say out loud. If they ask to jump straight to it, let them: paste prompt 01 first, then unit 05. Good output is a board with one lane per owner, every card carrying an owner and a date, and a printed list of any date set before the step it depends on.

Know these three facts and say them when they matter:
- Claude Data needs a data connection such as a warehouse or a connected app. A dashboard cannot use a sheet as its source, so a team with exports only gets the Sheets, Whiteboards and Meetings units.
- A sheet holds up to 100,000 cells and has no pivot tables yet. For full Excel, export it as .xlsx.
- Claude Meetings saves a transcript, not audio, and everyone on the call should agree to being transcribed.

Common problems and the fix:
- "I do not see Sheets or Whiteboard." On Enterprise an owner has to turn it on in Organization settings > Artifacts. On Team an owner may have turned it off. Ask the owner.
- "Claude Data asks me to connect something." That is correct. It needs a data connector. Prompt 04 helps them connect one, read only, on their own account, or they skip units 08 and 10 for now.
- "Claude will not join my call." Check the plan, that it is Google Meet, that an owner enabled Google Calendar and Meetings, and whether someone from outside the company is on the invite. By default Claude stays out of those calls.
- "The numbers do not tie." Good. Unit 10 is doing its job. Read the named break, fix the source, rerun.
Step 2 · run it on your data

Step 1 set it up. These 10 prompts do the work.

the vault

The 10 prompts

Grab the whole pack as one file, or tap any prompt below to copy it on its own. Placeholders that look like {{THIS}} get swapped for your own numbers — and if you ran Step 1, Claude fills them in for you.

One .md file · all 10 prompts, numbered, in order · nothing left out.
Click to copy
CLAUDE SUITE FOR FINANCE: SHARED RULES

1. WHO HAS ACCESS (checked against Anthropic's pages on 12 November 2026)
- Claude Meetings: beta, Team and Enterprise only, Google Meet only for now. Not listed for Pro, Max or Ultra. On Enterprise an owner must turn it on.
- Claude Sheets and Claude Whiteboards: beta on Pro, Max, Ultra, Team and Enterprise. On by default on Team. Off on Enterprise until an owner turns them on.
- Claude Data: generally available on the same five plans. It also needs a data connection. On Enterprise it turns on by default from 19 November 2026 unless an owner saved a setting.
- The free plan is not listed for any of the four. Treat them as paid plan features and check Anthropic's pages before relying on one.
Without Team or Enterprise, every unit still runs except the Meetings unit (09), which has its own fallback.
KEY FACT: a dashboard, doc or deck cannot use a Sheet as its data source, and a Sheet cannot pull live data.

2. WHERE TO RUN IT
All units run in a normal chat in the Claude app on the web or in Claude Desktop. Mobile is for viewing and light edits. Whether Claude Code can create these artifacts is NOT IN SOURCE: if a folder path or file listing is pasted, say so and stop.
Model: select Claude Opus 5.5 in the model picker. Use Claude Sonnet 5 only for a first bulk read of a very large export.
Privacy: your materials go only to your own Claude account, never to us. Use a Team or Enterprise plan, or turn model training off, before you add confidential numbers.
Limits, privacy exclusions and admin switches for each feature are in prompt 03. Anthropic's page wins over this pack.

3. BEFORE THE FIRST RUN
First ask what worries the human most about the next budget or forecast, and what they expect the numbers to show. Record the answer in their words. Then ask one at a time: {{PLAN}} (Pro, Max, Ultra, Team or Enterprise), {{MEETING_PLATFORM}}, {{ACTUALS_HOME}} (a warehouse, a connected app, or exports only), {{FISCAL_START}}, and {{SIGNOFF}}, the named person who approves anything that goes to the board. A missing item gets a plain default, labelled ASSUMED. Then print one availability line per feature: AVAILABLE, NEEDS AN OWNER STEP (name the step), NEEDS A DATA CONNECTION (Claude Data only), NOT ON THIS PLAN, or NOT IN SOURCE. In a sample run, treat the stated worry as given and do not ask.

4. STOP RATHER THAN GUESS
A missing document: ask once, name it, mark it OPEN. A stop blocks only the outputs that need it. Everything else still runs. If a feature is unavailable, say what it blocks and name the nearest thing that still works.

5. LABELS ON EVERY FIGURE AND DATE
REPORTED (source and page or cell) · CALCULATED (show the arithmetic) · ASSUMED (who set it) · MARKETING (a launch or vendor claim, including any time saved claim: a question to test, never a number) · NOT IN SOURCE.

6. THREE FLAG TAGS
- CHANGES THE DECISION: it reaches the output the board or the named person sees.
- WORTH A QUESTION: only if the answer could move a total by more than 1 percent or a date by more than 5 working days (ASSUMED until the human sets another).
- EXPLAINED BY CONTEXT: one line under "Checked, normal". Once closed, it stays closed.
Test every flag against the list below before you raise it. If an item fits two tags, the higher one wins. The 1 percent applies to the total the figure sits in.

7. LOOKS WRONG, IS NORMAL
PATTERN | WHY IT IS NORMAL | WHEN IT IS NOT
Sheets has no pivot tables, macros or Power Query | Documented beta limits | The model needs them. Then export to .xlsx and finish in Excel
The dashboard shows no forecast line | A dashboard cannot read a sheet | The forecast also sits in a table in the connected source
The Claude Data refresh time is hours old | It refreshes on open, not on a schedule | The board pack quotes the figure as current
Claude answers in the Meet chat in writing | That is the default | Everyone on the call should not see the answer
Claude will not join a call with an outside guest | It is the default, and an admin can change it | The owner intended it to join
Cash falls in a payroll week | Payroll lands in one week | The fall exceeds payroll plus the net of other lines
A whiteboard date was not pulled from the ledger | Whiteboards do not use connected app context | A person typed it wrong

8. STANCE
A neutral finance colleague who has run a budget cycle many times. Do not treat a launch claim as proof of a capability: test it on the human's own case.

9. HANDOFF (every unit ends with this)
Read (and not read) · Outputs · Checks: pass, fail or not run · Open items and the document needed · Next unit · Sign off needed from the named person.
Click to copy
Follow the shared rules in prompt 01. If prompt 01 is not in this chat, ask the human to paste it first. Run this sample before you use real data. Write your answer before reading the key. One input is missing on purpose, so you see the pack stop instead of guess.

SAMPLE (synthetic, every item cites its page)
Case A: Harbor Lane Supply, a finance team of 4 (intake notes p.1)
- Plan: Team. Meetings run on Google Meet. Actuals live in monthly CSV exports from the ledger. There is no data warehouse and no connected app (p.1).
- The controller, Dana Reyes, is a Team owner and signs off (p.1).
- The CFO says: "I want the board dashboard to show our forecast next to actuals, built from the forecast sheet." (p.1)
- The current model uses totals and sums only. It has no pivot tables and no macros (p.2).
- A colleague asks: "Will Claude say the answers out loud in the meeting?" (p.2)

Case B: a 3 week cash view for Harbor Lane Supply (cash file p.1)
Opening bank balance on the start date: 250,000.
Week 1: receipts 60,000. Supplier payments 65,000. Payroll 0.
Week 2: receipts 55,000. Supplier payments 58,000. Payroll 40,000.
Week 3: receipts 70,000. Supplier payments 52,000. Payroll 0.
A draft dashboard someone built in another tool from an AP report shows closing cash, end of week 3, as 260,000 (dashboard export p.1).
No taxes or debt service fall in these 3 weeks. No minimum cash level is set. The fiscal year starts 1 January. The worry in the controller's words: cash in week 2.
The AR ageing for weeks 4 to 13 was not provided.

WRITE
1. One availability line per feature.
2. The flags, each with one tag.
3. Closing cash for weeks 1 to 3, the lowest week, and the tie out of the dashboard figure.
4. The trap you avoided.
5. What is OPEN and what it blocks.

ANSWER KEY
Availability: Sheets AVAILABLE and Whiteboards AVAILABLE (Team, on by default). Meetings NEEDS AN OWNER STEP (Dana must enable Google Calendar first). Claude Data NEEDS A DATA CONNECTION, OPEN: the team has only CSV exports. Headroom to minimum cash: none set, so not calculated.
Closing cash: week 1 = 250,000 plus 60,000 minus 65,000 = 245,000. Week 2 = 245,000 plus 55,000 minus 58,000 minus 40,000 = 202,000. Week 3 = 202,000 plus 70,000 minus 52,000 = 220,000. Lowest week: week 2 at 202,000.
Tie out: dashboard 260,000 against sheet 220,000, difference 40,000, which equals the week 2 payroll. The AP report left payroll out, so do not use the dashboard figure until it includes payroll.
Flags:
1. No pivot tables in Sheets: EXPLAINED BY CONTEXT (the model uses totals and sums only).
2. Claude answers in writing in the Meet chat: EXPLAINED BY CONTEXT (the default). Speaking aloud: NOT IN SOURCE.
3. Cash fell in week 2 although receipts were 55,000: EXPLAINED BY CONTEXT (payroll week).
4. The 40,000 break between dashboard and sheet: CHANGES THE DECISION.
5. The CFO's dashboard from a forecast sheet: CHANGES THE DECISION (see the trap).
The trap: building a dashboard with the forecast "from the sheet" (the key fact in prompt 01). The right answer: forecast in the Sheet, dashboard of actuals from a connected source, or OPEN until one exists.
OPEN: the AR ageing for weeks 4 to 13. It blocks only those weeks' receipts. Weeks 1 to 3, the availability lines and the tie out still run.

A run that flags a normal item, builds a forecast dashboard from the sheet, or guesses the weeks 4 to 13 receipts has failed. If any figure does not match, find out why before you use real data.
Click to copy
Follow the shared rules in prompt 01. The human may never have opened these features. Walk them through ONE feature at a time in plain words, using only this guide. Anything not written here is NOT IN SOURCE: say you do not know and point them to Anthropic's page for that feature (prompt 04 lists them). Ask first which feature they want to start with.

KEY FACT, say it before they build anything: a dashboard cannot use a Sheet as its data source. The forecast stays in the Sheet. The dashboard shows actuals from the connected source.

CLAUDE SHEETS (a spreadsheet inside Claude)
Where: the Artifacts tab, then the Sheets tile. Or ask for a sheet in any chat.
First one: ask for what you need with your own numbers, for example "Make a hiring budget by team and quarter." Sheets start private to you. Click Share to bring people in.
Can do: tabs, formulas across tabs, formatting, dropdowns, checkboxes, charts, cell comments. You, your team and Claude edit the same cells, and each Claude edit is outlined. Version history. Imports CSV, Google Sheets and Excel. Exports .xlsx and CSV.
Cannot do: more than 100,000 cells, pivot tables, PDF export, macros, VBA, Power Query, live data from connected apps. Basic read only on mobile. The list of its 144 functions is unknown.
Admin switch: Organization settings > Artifacts. Plans and defaults: prompt 01. Not supported for HIPAA, CMEK or ZDR organizations.

CLAUDE WHITEBOARDS (a shared board)
Where: the Artifacts tab, then the tile labelled Whiteboard. Or ask in any chat.
First one: say "Start a whiteboard to plan our budget cycle." You can dictate with the microphone next to the send button, type, or sketch. Drag across part of the board to ask Claude about just that part.
Can do: draws shapes, arrows and sections, groups ideas, orders steps, turns a board into a timeline, suggests missing steps, works live with your team, turns the board into a doc or deck. Exports PNG.
Cannot do: it uses only what is on the board, not your connected apps. You cannot import a board from a file. Not available to HIPAA ready organizations.
Admin switch: Organization settings > Artifacts. Inviting people outside the organization needs Email invitations outside your organization turned on there.

CLAUDE DATA (dashboards, data stories, short videos from your data; was Claude Dashboards)
Where: the Artifacts tab, then Data. Or ask in any chat.
First one: connect your data first, either a warehouse (Amazon Redshift, BigQuery, ClickHouse, Databricks or Snowflake) or another app already connected to Claude, such as Salesforce. Then ask a plain question and name the table, for example "How does this quarter's revenue compare with last quarter's, by region?"
Can do: dashboards, Scrollyteller (a scrolling story), Visual short (a short video), Visual analysis (a written summary). Each chart shows its SQL query and when its data was last refreshed. Version history. Certified queries and live sharing are beta.
Cannot do: it does nothing without a data connection. Whether it accepts pasted or uploaded figures is unknown. It refreshes when someone opens it, not on a schedule.
Admin switch: Organization settings > Artifacts. Certified queries and live sharing have their own settings under Data.

CLAUDE MEETINGS (Claude joins a Google Meet call)
Where: Meetings in Claude or Claude Desktop.
First one: in Meetings, connect Google Calendar, select a meeting, click Add Claude. An owner must enable Google Calendar first. Auto join is under Settings > Capabilities > Meetings. Tell everyone on the call they are being transcribed.
Can do: joins as a visible participant, takes notes and action items, answers in writing in the Meet chat, and builds a doc or whiteboard when asked. Everything lands on the meeting page with the transcript.
Cannot do: it is not available to ZDR, HIPAA ready or CMEK organizations. It saves no audio. Whether it speaks aloud, and whether it edits an existing Sheet live, is unknown.
Admin switch: Organization settings > Products > Claude Meetings (beta). By default Claude stays out of calls with outside guests; an admin can allow them.

<trap>Answering from memory. These features are days old and their limits change. Also letting the human plan a dashboard built from a Sheet.</trap>
Stop: if the human asks about a plan, limit or feature this guide does not state, say it is unknown and name the Anthropic page to check. The rest of the guide still applies.
END with the handoff block from prompt 01. Next: prompt 05, the budget cycle on a Whiteboard, unless their first job is a different one.
Click to copy
<unit name="tools-setup">
Follow the shared rules in prompt 01.
What it does: checks the facts for each feature against Anthropic's pages, and helps connect the data source Claude Data needs, on the human's own account.
Use when: after prompt 02 passes. Skip the connection part without a warehouse or connected app.
Needs: {{PLAN}}, {{ACTUALS_HOME}}, and the names of the systems the human already uses to keep actuals.
Method:
1. Open these Anthropic pages and compare each fact in prompts 01 to 03 against them. Report any that changed. https://claude.com/resources/articles/sheets-whiteboards-data , https://support.claude.com/en/articles/18324300-get-started-with-claude-sheets , https://support.claude.com/en/articles/18324319-get-started-with-claude-whiteboards , https://support.claude.com/en/articles/17454700-get-started-with-claude-data , https://support.claude.com/en/articles/18325154-use-claude-meetings .
2. Ask the human to open the connectors area of their Claude and tell you what is connected. For the steps, read https://support.claude.com/en/articles/11176164 and follow it as written, never from memory.
3. For Claude Data, find the connector for the human's own warehouse or app (supported names are in prompt 03). Prefer an official publisher, read only access, sign in by OAuth, updated in the last 6 months.
4. Show one table: need, tool, publisher, official page URL, read only or read and write, reach, cost on the human's plan ("not stated" if absent), last update ("no date shown" if none).
5. For each tool the human approves, follow its official page and copy any command or URL word for word. If the page has no steps, mark it ASK VENDOR, write the one question for the human's admin, and use EXPORTS meanwhile.
6. Connect one at a time, then run one small read only test, for example one row count, and show the result.
Rules: the human's own accounts and credits only. Never ask for a key in chat. Start every connection read only.
<trap>A server that is not from the vendor, asks for write access or broad scopes, or sends numbers to a third party for processing. Leave it out, even when it is free.</trap>
Stop: no official page means do not install it. Mark that need EXPORTS and carry on. Without a data connection, unit 08 does not run and unit 10 runs in its reduced form.
Output: the changed rows, the connector table, what was installed, each test result, and the handoff. Next: the first working unit.
</unit>
Click to copy
<unit name="budget-cycle-whiteboard">
Follow the shared rules in prompt 01. If prompt 01 is not in this chat, ask the human to paste it first. If they decline, run anyway with these minimum rules: ask which Claude plan they are on, label every date and name the human typed REPORTED, label everything Claude adds ASSUMED, and end with the handoff.
What it does: turns the human's spoken or typed plan into a shared board with owners, dates and dependencies.
Use when: the steps, owners and dates of the budget cycle live in someone's head or in scattered emails. This is the best first unit and needs no files: a spoken plan is enough.
Needs: the board date, {{FISCAL_START}} (ask if not already given), the list of budget owners with their function, and last year's calendar if one exists, by document name.
Method:
1. Ask the human what worries them about this cycle and record it. Then ask for the board date and work backwards from it.
2. Check the human can open Whiteboards (Artifacts tab, Whiteboard tile). If not, name the owner step from prompt 01 and stop. Otherwise ask Claude for a whiteboard in the chat, for example "Start a whiteboard to plan our budget cycle." Let the human talk through the plan in their own words, by dictation or typing. Whiteboards works from what is on the board, so every date and name must be typed or pasted in by the human.
3. Lay out one lane per owner. Each card shows task, owner, due date, depends on and deliverable, and connects to the card it depends on.
4. Check the dates. For every dependency, calculate the working days between the two cards and show the arithmetic. Flag any card dated before the card it depends on.
5. Ask Claude what is missing between steps. Mark every card Claude adds ASSUMED until the owner confirms it.
6. Only after the named person signs off, ask for the board as a project plan doc.
<trap>A board that looks complete but has cards with no owner, or dates Claude filled in because it could not see last year's calendar. A date the human did not type is a date Claude made up.</trap>
Stop: no board date means lay out the cycle in weeks counted back from the board, labelled ASSUMED, and ask for the date. The lanes, the owners and the dependency check still run. An owner the human cannot name stays on the card as OWNER OPEN.
Output: the board, the cards with no owner, the dependency conflicts with arithmetic, then the handoff. Next: driver-forecast-sheet.
</unit>
Click to copy
<unit name="driver-forecast-sheet">
Follow the shared rules in prompt 01.
What it does: builds a forecast in Claude Sheets driven by a small set of inputs, and proves the drivers on last year first.
Use when: the team needs a forecast the owners can change, with every figure traceable to an input.
Needs: last year's actuals by month and by line (a CSV, an Excel file, a Google Sheets import, or pasted figures), the driver definitions, and the human's view of what changes next year.
Method:
1. Ask the human what they expect to change next year before you propose anything. Then propose 3 to 5 drivers per major line, with the reason for each, for example units times price, or headcount times loaded cost. The human chooses.
2. Import the actuals (Sheets reads CSV, Google Sheets and Excel files, or paste them). Build four tabs: Inputs, Actuals, Forecast and Checks. Every assumption sits in Inputs with its owner and the label ASSUMED. Forecast cells are formulas that point at Inputs. A typed number there is an error.
3. Back test. Run the drivers on last year's input values and compare to last year's Actuals, line by line. Print every line that differs by more than 2 percent (ASSUMED until the human sets another).
4. In Checks, show that the Forecast total equals the sum of its lines, with the arithmetic.
5. Count the cells used against the 100,000 cell limit. If monthly detail by line would pass it, cut the detail and say what was cut.
6. Confirm each function exists in Claude Sheets before relying on it (Anthropic gives a count of 144, not a list). Replace any that is missing and say so. Use no pivot tables.
<trap>A driver set that gives a sensible number for next year but cannot reproduce last year. Run the back test before the human sees the forecast, and put typed values in Inputs only.</trap>
Stop: no prior year actuals means build the four tabs with an empty Actuals tab and no forecast figures. Inputs and Checks still run.
Output: the sheet, the back test table, the list of ASSUMED inputs with owners, then the handoff. Next: cash-13-week-sheet.
</unit>
Click to copy
<unit name="cash-13-week-sheet">
Follow the shared rules in prompt 01.
What it does: builds a 13 week cash view in Claude Sheets showing the lowest cash week and the headroom to the minimum cash level.
Use when: the team needs to know the first week cash is tight.
Needs: the bank balance on the start date by account (a bank statement), the AR ageing or expected collections by week, the AP schedule, the payroll calendar, tax and debt service dates, and the minimum cash level if any.
Method:
1. Ask what worries them first: a covenant, a payroll week, a large supplier payment. Record it.
2. Build a tab with weeks 1 to 13 as columns with start dates. Rows: receipts, supplier payments, payroll, taxes, debt service, other, opening cash, net movement, closing cash.
3. Closing cash = opening cash plus receipts minus payments of every kind. Each week's opening cash is a formula pointing at the prior week's closing cash.
4. Receipts come from the AR ageing, by invoice due date plus the days late the customer has paid in the past. If no payment history is given, use zero days late and label it ASSUMED.
5. Print the lowest closing cash, its week, the headroom to the minimum cash level, and the first week below it.
6. Put every input in an Inputs tab. Nothing is typed into a weekly cell.
<trap>Payroll, taxes, annual insurance or debt service that sit outside the AP schedule, so closing cash looks healthy and is not. Also an opening balance from the ledger when the bank balance differs because of unpresented payments. State which one you used.</trap>
Stop: no AR ageing means build the payments side for all 13 weeks, take receipts only for weeks the human can support, mark the rest OPEN and do not guess collections. If payments data is also missing, build the layout and formulas with empty inputs and mark every week OPEN.
Output: the sheet, the lowest week table, the OPEN receipts weeks, then the handoff. Next: board-dashboard-data.
</unit>
Click to copy
<unit name="board-dashboard-data">
Follow the shared rules in prompt 01.
What it does: builds a board dashboard and a short data story in Claude Data from the human's connected source, and checks the query behind every chart.
Use when: the board wants actuals on one page.
Needs: a warehouse or app already connected to Claude, the table that holds actuals, the metric definitions, and certified queries if the data team has any.
Method:
1. Confirm the connection exists. If {{ACTUALS_HOME}} is exports only, stop here.
2. State what the dashboard will and will not show: actuals from the connected source. The forecast stays in the Sheet (key fact, prompt 01) unless it also sits in a table in the connected source, in which case the dashboard may show it from there, labelled.
3. Ask for the dashboard, naming the table and the question, for example this quarter's revenue against last quarter's by region. Open each chart's query. List one row per chart: table, filters, period, the number shown, and the last refreshed time.
4. Ask for a data story from the same metrics. Check every sentence's number against the chart's query before sharing. Make a video only after the text is checked.
5. Mark a chart CERTIFIED only if it carries the certified badge (certified queries are beta). Otherwise UNCERTIFIED.
6. Live sharing is beta and runs on the sharer's access: anyone it is shared with sees results from the sharer's source, even without access to it. Keep it off until the named person has seen the dashboard.
<trap>A chart not tied to a query on the connected source, for example built from numbers typed into the chat. Also treating a figure as current when it refreshed hours ago: Claude Data refreshes on open.</trap>
Stop: no data connection means this unit does not run. It blocks the dashboard, the data story and the dashboard side of unit 10. Everything else still runs.
Output: the chart table, the checked data story, the sharing settings, then the handoff. Next: budget-review-meeting.
</unit>
Click to copy
<unit name="budget-review-meeting">
Follow the shared rules in prompt 01.
What it does: prepares and runs a budget review on Google Meet with Claude as a visible participant, and gets the decisions recorded, read back and applied to the Sheet.
Use when: the owners meet to agree changes to the plan, forecast or cash view.
Needs: Team or Enterprise, Google Meet, an owner who has enabled Claude Meetings and Google Calendar, the forecast Sheet, the unit 05 whiteboard, and an agenda.
Method:
1. Before the call, check the setup against https://support.claude.com/en/articles/18325154 . In Claude, go to Meetings, connect Google Calendar, select the meeting and choose Add Claude. Everyone on the call must be fine with being transcribed.
2. Check the guest setting. By default Claude stays out of calls with someone from outside the organization, and an admin can allow it. Where allowed, Claude takes notes and does not reply in the chat. Plan the review as internal.
3. At the start, test a live Sheet edit: Anthropic shows docs and whiteboards in a meeting, not a Sheet edit. Ask Claude to change one throwaway cell in a copy of the Sheet. If it cannot, record decisions on a whiteboard and update the Sheet after the call.
4. During the call, ask Claude to record each decision as one row: item, old figure, new figure, owner, who decided, and to read all rows back before the call ends. Anything Claude posts in the Meet chat is visible to all.
5. After the call, open the meeting page (notes, action items, transcript, everything Claude made; audio is not saved). Check each decision row against the transcript, then apply the changes to the Sheet.
6. Claude asks permission before it uses a connected tool, and those answers count toward usage. Say so before the call.
<trap>Believing Claude changed the Sheet because it said so. Read the cells. A figure said aloud may be transcribed wrongly, so a decision counts only after it is read back and the named person confirms it.</trap>
Stop: if the plan is not Team or Enterprise, the organization has ZDR, HIPAA ready or CMEK, the call is not on Google Meet, or no owner has enabled Meetings, this unit does not run. Run the review in a normal chat instead: agenda from the unit 05 whiteboard, each decision typed as a row, Claude reads all rows back, then apply them to the Sheet and check the cells. Every other unit still runs.
Output: the decision rows, the cells changed with before and after, the action items, then the handoff. Next: dashboard-sheet-tie-out.
</unit>
Click to copy
<unit name="dashboard-sheet-tie-out">
Follow the shared rules in prompt 01.
What it does: re-derives the key figures from the ledger, then checks the Sheet and the dashboard against it and each other.
Use when: before anything goes to the board.
Needs: the ledger export for the period, the Sheet, the chart table from unit 08, and the key figures the human names (default: total revenue, total operating cost, closing cash at the stated date).
Method:
1. Before you read the Sheet or the dashboard, recompute each key figure from the ledger export. Show the arithmetic. Label each CALCULATED.
2. Compare each recomputed figure to the Sheet, then to the dashboard chart. A tie means the same figure at the unit shown. For every break, show the difference and the arithmetic.
3. Test each break against the normal list in prompt 01 (cut off dates, rounding, dashboard refresh time against the Sheet's date). Close what is explained and tag the rest.
4. Print TIES or BREAK for each figure. While any break tagged CHANGES THE DECISION is open, board use of that figure is blocked. Keep running the other checks so nothing is hidden behind the block.
5. Print the default if the human does nothing: the Sheet figures go to the board labelled CALCULATED and the dashboard figures are held out.
<trap>The Sheet and the dashboard read the same extract and are wrong in the same direction, so they agree. That is why step 1 starts from the ledger.</trap>
Stop: no ledger export means the Sheet to dashboard comparison still runs, labelled TIES TO EACH OTHER, NOT TO THE LEDGER. No dashboard means the ledger to Sheet check still runs.
Output: the tie out table, the open breaks with tags, the default, then the handoff. Next: sign off by the named person.
</unit>

From a working build to an operating system.

For finance teams: find and use Claude Sheets, Whiteboards, Data and Meetings for the budget cycle, a driver forecast, 13 week cash and the board dashboard, with a tie out to the ledger.

Path A · free

Run the build independently

The setup rail and every prompt above are free and stay free. The cost is your time, and the risk of wiring it wrong on live data.

Back to the prompts
Path B · done with you

Connect it to your operation

These are Anthropic features you switch on and run yourself. Consultance wires the last mile in your environment: a read only connection from your warehouse so the dashboard reads governed actuals, certified metric definitions, review gates before the board, and the owner settings your admin needs. Reply wire it for a 30 minute slot.

Book a build call
data safety

Before you use live numbers

  • • Run last quarter's numbers first. Live data is not a test bed.
  • • Nothing here uploads to us. It runs in your own Claude account, on your own machine.
  • • A named human reviews and signs every output before it reaches a board, lender, or client.
  • • Mask account numbers and names to the minimum the task needs.
the fine print

Straight answers on ownership

Prompt set authored by consultance.ai. Nothing is hosted by us: your numbers go only to your own Claude account, under your own Claude plan's terms, never to us. Use a Team or Enterprise plan, or turn off model training, before loading confidential figures. Features, plans and limits are Anthropic's and change often, and the pack tells you to check Anthropic's pages. Not accounting, audit, tax or investment advice, and a named person signs off every figure.

Bring this workflow into your operation.

Book a build callBack to the library

Want this wired into your stack instead of running it yourself? That is our AI deal desk and finance automation service.

Newsletter

New builds, as they ship

One short note when a new build ships: what it does, who it is for, and how to run it in your own Claude.

Free. Three short welcome emails over your first three days, then at most one email a week. Unsubscribe in one click.

More like this

Related builds

All builds
Finance and data
Finance and data

Finance Team Library Guide

For in house finance teams: one guide to 51 free packs, mapped to close, cash, spend, planning, board, tax, payroll and AI rollout, with the order to run them.

Setup guide
Finance and data
Finance and data

The FP&A Agent Stack

For CFOs, controllers, and FP&A leads: the map of ten agent desks that run the planning cycle inside your own Claude, from the overnight analyst roster and the rolling forecast to variance, cash, close, reporting, the board pack, and the statement rec, plus the order to install them in.

Setup guide
Finance and data
Finance and data

Allocator Excel Desk

For allocators, hedge PMs, and PE operating partners: demo the desk on a public sample you can show in a meeting, then run the same gate on the internal workbook that stays on your machine. A local Python desk (openpyxl + pandas) writes DPI, TVPI, IRR, Kaplan Schoar PME, and a PASS or BLOCK back into the file. Hedge mode flags name concentration. The browser tool is the public demo only. Not a Burgiss or Aladdin replacement.

Setup guide
Get the free kitBook a call

From independent build to integrated workflow

What you have

The free version above. It runs this read on a file you paste in, in your own Claude. Yours to keep.

What we install

The same read, wired into your cfo and back office stack, running the second a report lands. Nothing to upload, no new login.

See the installed version

Share with your workflow owner.

Bring the relevant team into the conversation with the build and setup guide.

Forward by email
in one line

What is Claude Sheets, Whiteboards and Data for Finance Teams?

Claude Sheets, Whiteboards and Data for Finance Teams is a finance and data build in the consultance.ai AI Build Library. For finance teams: find and use Claude Sheets, Whiteboards, Data and Meetings for the budget cycle, a driver forecast, 13 week cash and the board dashboard, with a tie out to the ledger. It fits CFOs, controllers and FP&A leads on a Claude Pro, Max, Ultra, Team or Enterprise plan who have heard about the new Claude features and want to know which ones their plan allows and how to use them on the budget. Setup difficulty is Easy, with 5 plain-English steps.

What does Claude Sheets, Whiteboards and Data for Finance Teams do?

For finance teams: find and use Claude Sheets, Whiteboards, Data and Meetings for the budget cycle, a driver forecast, 13 week cash and the board dashboard, with a tie out to the ledger.

Who is Claude Sheets, Whiteboards and Data for Finance Teams for?

It fits CFOs, controllers and FP&A leads on a Claude Pro, Max, Ultra, Team or Enterprise plan who have heard about the new Claude features and want to know which ones their plan allows and how to use them on the budget.

How hard is Claude Sheets, Whiteboards and Data for Finance Teams to set up?

Easy to set up — one guided setup instruction covering 5 plain-English steps, plus 10 ready-to-run prompts on the resource page.

How would consultance.ai build this out?

These are Anthropic features you switch on and run yourself. Consultance wires the last mile in your environment: a read only connection from your warehouse so the dashboard reads governed actuals, certified metric definitions, review gates before the board, and the owner settings your admin needs. Reply wire it for a 30 minute slot.

What are the licensing terms?

Prompt set authored by consultance.ai. Nothing is hosted by us: your numbers go only to your own Claude account, under your own Claude plan's terms, never to us. Use a Team or Enterprise plan, or turn off model training, before loading confidential figures. Features, plans and limits are Anthropic's and change often, and the pack tells you to check Anthropic's pages. Not accounting, audit, tax or investment advice, and a named person signs off every figure.

Connect this build to your workflow.

Discuss how Claude Sheets, Whiteboards and Data for Finance Teams fits your systems, controls, and team. We define the integration, review process, and deployment requirements with you.

This build comes from our AI consulting and AI implementation practice — see the full AI in finance guide and how we work with CFO teams.

Book an audit