Mike's Checks
Checks

Mike's Checks/meta-muse-spark-1.3/05 showrunner

05 showrunner

meta-muse-spark-1.3OpenCodehigh effortrun 8 Sep 202651,699 tokens

Compare models
16/16
checks passed
100%
05 showrunner · meta-muse-spark-1.3100%
▸Instructions — the case's current instructions; none were saved with this result

Hi, I need to create a modified Claude code for beginners session, like run sheet basically, run of show, that has the following modifications. So I'm going to upload a few different examples of things I've done in the past, just so you have them to work from. But basically, there's a few modifications for this one which make it unique. One is that this is only using Claude Cowork, so we should not be using Claude Code, either in the desktop app or in the terminal, because this team is non-technical. The team are all working in financial services slash accounting for a education company. So what kind of workplace education company called Brightmoor Education. So the sorry that spelled b the uh team is very busy and um doesn have a lot of like time to explore speculative use cases of ai they want this to be incredibly practical and based on their existing workflows. I'm uploading some transcripts from a call we had with them and also just some notes on what I think could work. Specifically, they're interested in... creating the dashboards and specifically they mean just making the information look visibly better like like creating a html dashboard in a nice kind of style from from existing say data that contained in Excel They also interested in doing financial analysis So more of a, you know, like, I guess, a cloud coworker would use Python to do the analysis and then maybe output the analysis in Excel or the input would be in Excel. And then the third thing was like making a PowerPoint where they update the PowerPoint template automatically using Claude Cowork so that they can do kind of weekly or monthly presentations on the data. So that's the goal, is to kind of make it very focused on that. I still want to keep the way that we introduce Claude Cowork, where we explain to it that the difference between Claude Cowork and ChatGPT is that it has access to the file system. It can run code. I think that important But maybe the ASCII image idea that I normally do where I ask it to create an ASCII image based on my location isn like that useful I think we need to make it more practical for them So maybe we use an Excel implementation instead. The assumption is... oh, so they use Snowflake, by the way, for all their data, and that is powered by NetSuite, which is our accounting software. I guess that's by Oracle. So when we create synthetic data, we'll be, for these tasks that we're going to demo, we'll be focused specifically on solving that, solving those problems. So like the shape of the tasks, the shape of the data should be like NetSuite shaped, if that makes sense. So make it very tight run down, make it very condensed and practical. No real like wider theory or like, hey, we're excited about AI things. I think these guys just want to get stuff done.
So yeah, let me know what the run should be.

some very specific asks in here

  • definitely non-technical (claude cowork only + claude in excel)
  • core focus on accounting/finance function
  • use cases to demo: creating a dashboard, doing financial analysis, making a powerpoint

heavy focus on practical stuff rather than theory - they're just super busy and want to get stuff done, have heard from friends at other companies you can do it with claude

For accounting: put financials together, doing reconciliations, identifying trends in data sets, putting together repeatable compliance reports For planning & forecast: build financial models (make assumptions and pull & clean data from snowflake)create dashboards, and publish to executivesAutomate charts on slides

Answer

ANSWER.md the model's output · 11.7 KBdownload

Brightmoor Education — Claude Cowork for Finance: Run of Show

Audience: 20-30 Finance (Accounting, Payroll, Bursar + Strategic Finance) — non-technical, very busy, beginner (5-10% intermediate)
Tool: Claude Cowork ONLY. No Claude Code Desktop / Terminal. + Claude in Excel where noted.
Duration: 3.5 hr delivery in 4 hr block (30-min buffer for setup overrun)
Goal: Each attendee leaves having done 3 repeatable workflows on NetSuite-shaped data: (1) Excel → HTML dashboard, (2) financial analysis / reconciliation, (3) Excel → PowerPoint monthly update

Assumptions (from call notes)

  • Data source of truth: NetSuite (Oracle) → Snowflake. Live Snowflake connection is OPTIONAL for session — we use local synthetic files to avoid IT friction (2-hr provisioning noted).
  • Monthly pain: ~15-page PowerPoint from Excel export + commentary/charts = ~8 hrs/person. Dashboard pain: Excel formatting is painful; want clean modern view for senior leadership.
  • Dana/Strategic Finance already do: data cleaning in Claude, basic dashboards, some modeling in Claude + Claude in Excel. Rest are rookies for financial analysis.
  • No volunteer TAs (team too busy). No theory — practical only.
  • NDA / provisioning of Claude access to be confirmed pre-session. Invoice headcount day-before.

Non-goals (say this upfront)

  • No Claude Code, no terminal, no GitHub, no deployment, no API/MCP. No speculative AI strategy.
  • We will NOT leave with a perfect one-shot template. We WILL leave with a repeatable process + checked outputs.

Pre-work / Facilitator checklist (Mon–Thu)

Monday – send agenda + pre-work email (5 min for attendee):

  1. Confirm Claude Cowork access for all 20-30 attendees. Test login.
  2. Ask attendees to bring laptop + have Excel + PowerPoint desktop available. No installs needed beyond Cowork.
  3. Optional: Snowflake access — explicitly marked OPTIONAL. Session works without it.
  4. Send synthetic data pack download link (4 files below). Also preload in Cowork shared folder if org supports it.

Synthetic Data Pack — NetSuite-shaped (prepare Tue, QA Wed):
All fake numbers, real shapes. Campuses as left-hand dimension (per Dana demo). Dates: FY25 + FY26 YTD through Mar.

  1. BE_NetSuite_TrialBalance_FY25-FY26YTD.xlsx — Sheets: GL_Detail (JournalID, Date, Period, Subsidiary/Campus, Dept, Account_No, Account_Name, Account_Type, Debit, Credit, Memo, Source=NetSuite), CoA (chart of accounts), README (period close status)
    • Include intentional mess: duplicates, unmapped Dept 999-TBD, intercompany imbalance in one period, negative payroll accrual.
  2. BE_Campus_Enrollment_Revenue_Monthly.xlsx — Enroll (Campus, Program, Month, Headcount, Starts, Withdrawals), Revenue (Campus, Month, Tuition_Billed, Discounts, Net_Revenue) — 5 campuses: e.g., Bayview, Crosstown, Ideal Info Center, Northgate, Online.
  3. BE_Monthly_Close_Package_MAR.xlsx — The "ugly" starting point: 3 raw tables (P&L by campus, Opex variance, Cash recon) with merged cells, inconsistent formatting — this is Build 1 + 2 input.
  4. BE_Monthly_Deck_TEMPLATE.pptx — 6-slide template (not 15, to fit time): 1 Cover, 2 KPI tiles, 3 Revenue by Campus bar, 4 Opex variance table, 5 Enrollment trend line, 6 Commentary/Next steps. Branded with Brightmoor placeholder colors/fonts. Notes section has source cell references.

Keep all files <5MB, no external links, no macros.

Room setup:

  • Projector + screen share. One facilitator drives, one floats chat/Q&A.
  • Cowork folder structure demo: Brightmoor-Finance/inputs/, outputs/, archive/ — tell attendees to mirror this.
  • Timer visible. No breakout TA rooms — use neighbor-pair share only (team too busy for formal TA role).

Core framing to keep (5 min, plain English)

Difference between ChatGPT/Gemini chat and Claude Cowork: chat only talks about your file. Cowork sees your folder, reads Excel, runs Python to do the math, and writes new files back — dashboard HTML, cleaned Excel, updated PowerPoint. You check the numbers before you send to leadership.

Do NOT do ASCII-map icebreaker. Replace with Excel starter (below) — immediately practical.

Run of Show — 3.5 hr (210 min)

# Time Block What happens Facilitator script / prompt Output
0 0:00-0:15 (15) Welcome + Cowork in 3 ideas Why now for Finance: put financials together faster, reconciliations, trends, repeatable compliance reports; FP&A: clean Snowflake pulls, models, dashboards, exec decks. 3 ideas: (1) Cowork sees files, (2) runs code for math, (3) writes files you keep. Trust rule: AI drafts, you reconcile. Live demo (2 min): point Cowork at empty folder, show it listing files. Say: "If it can't see the file, it can't help — put it in the folder first." Everyone has Cowork open + data pack in folder
1 0:15-0:30 (15) Starter: Fix my NetSuite export Everyone does same tiny win. No Plan Mode jargon — use: Tell → Check → Fix. Copy-paste prompt: In BE_Monthly_Close_Package_MAR.xlsx, sheet P&L_by_Campus: unmerge cells, fix headers in row 1, add Totals that tie to GL_Detail in BE_NetSuite_TrialBalance file, and save as outputs/MAR_cleaned.xlsx. Show me any rows that don't tie, don't overwrite my input. outputs/MAR_cleaned.xlsx + tie-out list
2 0:30-1:10 (40) Build 1: Excel → HTML Dashboard (for senior leadership) Guided, all together. Input: MAR_cleaned.xlsx. Output: single-file dashboard_mar.html with KPI tiles (Net Revenue, Opex, Headcount), revenue-by-campus bar, enrollment trend, variance callouts. Brightmoor clean style. Then Tune: horizontal bar, filter by campus. Prompt 1: Using outputs/MAR_cleaned.xlsx, build a single-file HTML dashboard (dashboard_mar.html) for senior leadership: KPI tiles at top, revenue by campus bar chart, enrollment trend line, top 3 variances vs budget. Clean modern style, Brightmoor blue/gray, readable on laptop. No external dependencies. Data stays as displayed values — show source sheet + month in footer. Tune prompts (do live): Change campus chart to horizontal bar / Add a Campus filter dropdown / Make it printable to PDF on one page dashboard_mar.html open in browser. Publish note: save to shared drive / internal site — no external login needed today; IT publishes later
3 1:10-1:50 (40) Build 2: Financial analysis + reconciliation (Python does math, Excel keeps proof) Same files. Three tasks, do in order: (a) tie-out/recon, (b) variance & trend, (c) write-back to Excel with formulas visible. Emphasize verification. Prompt 2: Using BE_NetSuite_TrialBalance_FY25-FY26YTD.xlsx GL_Detail + BE_Campus_Enrollment_Revenue_Monthly.xlsx: (1) Find duplicates, unmapped Dept 999-TBD, and intercompany imbalance by period. (2) Compute Mar YTD Net Revenue, Opex vs Budget variance by campus, and 3-month enrollment trend. Use Python, then write outputs/MAR_analysis.xlsx with sheets: Recon_Flags, Variance_by_Campus, Trend, Method (list every formula/assumption). Keep all $ ties to GL_Detail totals — flag if off by >$1. Follow-up: Explain the top 2 variances in plain English for a controller, 3 bullets each, with Campus + $ + likely driver from Memo/Dept. outputs/MAR_analysis.xlsx (4 sheets) + 6-bullet commentary to paste into deck
— 1:50-2:00 (10) Break — — —
4 2:00-2:45 (45) Build 3: Excel → PowerPoint monthly update (the 8-hour saver) Guided. Input: MAR_analysis.xlsx + BE_Monthly_Deck_TEMPLATE.pptx. Output: BE_Monthly_Deck_MAR_DRAFT.pptx. Teach mirror-chart + repeatable update. Prompt 3: Open BE_Monthly_Deck_TEMPLATE.pptx and outputs/MAR_analysis.xlsx. Create outputs/BE_Monthly_Deck_MAR_DRAFT.pptx: keep template fonts/colors/layout. Update Slide 3 bar with Mar Net Revenue by Campus, Slide 4 table with Variance_by_Campus, Slide 5 line with enrollment trend. Add slide numbers + footer "Source: MAR_analysis.xlsx, Mar FY26". Do NOT invent numbers — leave placeholder text [CHECK] if a value is missing. List every slide you changed in chat. Then: Mirror Slide 3 style to make 2 more charts: Opex vs Budget by campus, Discount rate by campus. Match colors/fonts exactly. Draft deck + change log. Show how next month = swap input file + rerun same prompt
5 2:45-3:15 (30) Open apply: your close / your campus Attendees pick ONE: (a) their own ugly Excel → dashboard tweak, (b) recon on own export (redacted), or (c) template their own 1-slide update. Neighbor-pair share at 3:10. Float support. No new concepts. Provide 3 one-line prompts on slide. Rule: work on copy in outputs/, never original. 1 file improved per person
6 3:15-3:30 (15) Verify + What's next Trust checklist (below). Claude in Excel 1-min demo: =CLAUDE("summarize variance in plain English", range) alternative when Cowork not available. Close: repeatable folder + prompt library. Handout: 1-page Prompt Library + Verification Checklist (print/slide). Homework: run April close with same 3 prompts, bring one failure to office hours. Checklist + next steps

Total: 15+15+40+40+10+45+30+15 = 210 min.

If time runs short

Cut Build 2 commentary bullets (keep Recon_Flags + Variance). Cut mirror-chart second chart. Never cut verification (Block 6).

Troubleshooting — ASK Ladder (Cowork wording for non-technical)

  1. Ask to fix: That's not right — [what you see] vs [what you expected]. Fix only that and show me the file you changed.
  2. Simplify: Explain what you just did like I'm new to Excel. Show the formula.
  3. Start fresh: Duplicate prompt in NEW Cowork session with only the 2 files it needs. Cluttered context = wrong numbers.

Common failures + fixes:

  • Cowork invents numbers → add Do NOT invent numbers, use [CHECK] + require Method sheet.
  • HTML chart wrong → ask for horizontal bar + show data table under chart to verify.
  • PPT layout breaks → Keep template master, only replace chart data and text boxes, don't resize.
  • File not found → move file into Cowork folder first, then retry same prompt.

Verification Checklist (teach explicitly, 5 min in Block 6)

  1. Totals tie: outputs/* totals = GL_Detail Debit-Credit totals. Off by >$1 = stop.
  2. Method sheet exists: every calc lists source sheet + period + assumption.
  3. No hidden edits: input files untouched in inputs/, all work in outputs/.
  4. Human signs: controller initials + date before deck goes to leadership.
  5. Snowflake note: if you later connect Snowflake, same prompts work — just point to fresh export. No live connector needed today.

Prompt Library (give as slide/handout)

  • Clean: Clean [file] headers/totals, save copy to outputs/, list untied rows. Don't overwrite input.
  • Dashboard: Build single-file HTML dashboard from [cleaned file] for senior leadership, show source + month, printable.
  • Reconcile: Flag duplicates/unmapped/intercompany by period in [trial balance], write Recon_Flags sheet.
  • Variance: Compute YTD vs Budget variance by campus, top 3 drivers, plain-English bullets with $ + Campus.
  • Deck update: Update template [pptx] charts/tables from [analysis xlsx], keep style, mark missing as [CHECK], list slides changed.

Follow-up

  • 45-min office hours following week: bring April data, debug live.
  • Optional advanced (for Dana/Greg + 1-2 others): Snowflake connector setup with IT + Claude in Excel for repeatable compliance reports. Not in main session.

What changed vs Nvalo run

  • Removed: Claude Code Desktop/Terminal, GitHub, ASCII icebreaker, vacation planner, PM tools, deployment, Q&A with CEO.
  • Added: NetSuite-shaped synthetic pack, Excel-starter, HTML dashboard styling + print/filter tune, Python-backed recon/variance with Method sheet, PPT template mirror + monthly rerun, verification checklist, Claude in Excel bridge.
  • Tone: no "builders/extensions" theory. Language: folder, Excel, PowerPoint, tie-out, close.

Checks

16/16
Script checks 4/4answered by a program
pass

Run of show existsexists

a run-of-show document was produced

pass

Coherent scheduleformat

it parses as a run of show: one coherent schedule, no segment that ends before it starts, >=5 time-ranged rows carrying activity text (md table rows, list items, bold headings "(9:00-9:15)", single-clock table rows, or plain lines all count)

pass

Five-minute time markstime-grain

segment boundaries sit on 5-minute marks, at most one off-grid -- a schedule planned at :02/:08/:17 is arithmetic, not a plan

pass

Timings add uptimings-tie-out

per schedule, segments are contiguous: no overlaps, no gap over 30 min, >=85% of the span scheduled, and a span between 45 and 600 minutes

Judge checks 12/12judged by Claude
pass

Q1Uses Cowork throughout

Judge's reasoning

Explicit non-goal: "No Claude Code, no terminal, no GitHub, no deployment, no API/MCP"; all six blocks are Cowork plus a Claude in Excel bridge in Block 6.

▸Rubric

Cowork-only. Do all attendee-facing steps stay inside Claude Cowork (and Claude in Excel)? FAIL if any step the attendees are asked to do involves Claude Code, the terminal/CLI, git or GitHub, or installing developer tooling. The prompt: "this is only using Claude Cowork, so we should not be using Claude Code, either in the desktop app or in the terminal, because this team is non-technical."

pass

Q2Dashboard exercise

Judge's reasoning

Block 2 (0:30-1:10, 40 min) has attendees build `dashboard_mar.html` from `MAR_cleaned.xlsx` with a full paste-ready prompt plus tune prompts.

▸Rubric

Dashboard use case. Is there a hands-on segment where attendees build a dashboard from existing spreadsheet data? FAIL if dashboards are only mentioned, described or promised for later rather than built in the session.

pass

Q3Financial analysis exercise

Judge's reasoning

Block 3 (1:10-1:50, 40 min) is a separate recon/variance build: "Use Python, then write outputs/MAR_analysis.xlsx with sheets: Recon_Flags, Variance_by_Campus, Trend, Method".

▸Rubric

Financial analysis use case. Is there a hands-on segment where attendees do financial analysis (Claude writing/running code over their numbers, e.g. Excel in, analysis out)? FAIL if absent or folded into the dashboard segment as a passing remark.

pass

Q4PowerPoint exercise

Judge's reasoning

Block 4 (2:00-2:45, 45 min) is guided hands-on: attendees update `BE_Monthly_Deck_TEMPLATE.pptx` from `MAR_analysis.xlsx` and learn the mirror-chart rerun for next month.

▸Rubric

PowerPoint use case. Is there a hands-on segment where attendees update a PowerPoint template/deck from the data — the client's monthly reporting pack? FAIL if absent, or if it is only a demo the facilitator drives while attendees watch.

pass

Q5Relevant business data

Judge's reasoning

Data pack is NetSuite-shaped — `GL_Detail` (JournalID, Subsidiary/Campus, Account_No, Debit, Credit, Source=NetSuite), campus revenue/enrollment, Opex vs Budget variance, FY25 + FY26 YTD, with seeded intercompany imbalance and payroll accrual.

▸Rubric

NetSuite/Snowflake-shaped data. Is the synthetic data used in the exercises shaped like this client's actual data — NetSuite/Snowflake-style finance records (GL export, revenue and budget vs actual, EBITDA build-up, payroll vs non-payroll, campus/entity breakdown, monthly and year-to-date columns)? FAIL if the exercises use generic sample data (a demo CSV, sales widgets, made-up SaaS metrics) or leave the data unspecified. The prompt: "the shape of the tasks, the shape of the data should be like NetSuite shaped."

pass

Q6No ASCII icebreaker

Judge's reasoning

"Do NOT do ASCII-map icebreaker. Replace with Excel starter" — opening exercise is "Fix my NetSuite export" (unmerge, fix headers, tie totals to GL_Detail).

▸Rubric

No ASCII icebreaker. Is the opening hands-on exercise a practical finance/Excel task? FAIL if the run of show keeps the ASCII-image-of-your-location icebreaker or substitutes another whimsical non-work exercise. The prompt: "maybe the ASCII image idea that I normally do where I ask it to create an ASCII image based on my location isn['t] like that useful... maybe we use an Excel implementation instead."

pass

Q7Explains Cowork versus ChatGPT

Judge's reasoning

Core framing kept: "chat only *talks* about your file. Cowork *sees your folder, reads Excel, runs Python to do the math, and writes new files back*".

▸Rubric

Cowork vs ChatGPT explained. Does the run of show keep the explanation of how Cowork differs from ChatGPT — that it has access to the file system and can run code? FAIL if that framing is dropped. The prompt: "I still want to keep the way that we introduce Claude Cowork, where we explain to it that the difference between Claude Cowork and ChatGPT is that it has access to the file system. It can run code. I think that important."

pass

Q8No theory or hype

Judge's reasoning

Every block maps to a finance task (clean export, dashboard, recon/variance, monthly deck, own close, verification); non-goals bar "speculative AI strategy" and the change log removes theory language.

▸Rubric

No theory, no AI cheerleading. Is every segment tied to a task these people do at work? FAIL if any segment is devoted to AI industry context, the future of work, model capabilities, prompt-engineering theory, or excitement-building. The prompt: "make it very tight run down, make it very condensed and practical. No real like wider theory or like, hey, we're excited about AI things. I think these guys just want to get stuff done."

pass

Q9Mostly hands-on work

Judge's reasoning

Hands-on blocks total 170 of 210 min (15+40+40+45+30), leaving only welcome and wrap-up as non-build.

▸Rubric

Mostly hands-on. Is at least half the scheduled time attendees working with Claude themselves? FAIL if presentation, discussion and Q&A segments outweigh build segments. From the call: "we try and make at least 50% of it them actually you know working with us to do some of these tasks."

pass

Q10Realistic segment timings

Judge's reasoning

Builds get 40/40/45 min each — none is a 15-minute end-to-end build — and setup is handled via Monday login test, pre-shared data pack, a floating second facilitator, and an explicit 30-min buffer for setup overrun.

▸Rubric

Per-segment timings realistic. Could a room of 20–30 beginners actually finish each segment in the time allotted? FAIL if any build segment is implausibly short (a dashboard, an analysis or a deck refresh built end to end in ~15 minutes), or if setup/handholding time for first-time users is ignored.

pass

Q11Fits the booked time

Judge's reasoning

"Total: 15+15+40+40+10+45+30+15 = 210 min" — 3.5 hr delivered inside the stated 4-hr block.

▸Rubric

Total length fits the booked slot. Does the session run about three and a half hours, inside the four-hour block? FAIL if the total is materially shorter or longer (under ~3 hours or over ~4 hours) without the run of show explaining the change. From the call notes: "we have a four hour block for next week and the session itself is three and a half hours."

pass

Q12Ready to run

Judge's reasoning

Each block names input/output files, gives verbatim copy-paste prompts and tune follow-ups, plus a data-pack spec, prompt library, verification checklist, troubleshooting ladder and cut-list if time runs short.

▸Rubric

Usable as a run of show. Could a facilitator run the session from this document alone? FAIL if segments are titles without content — no prompts to paste, no data files named, no statement of what attendees produce — so the facilitator would still have to design the session.