How to produce executive business reviews that make hospitality leaders feel understood, show measurable impact, and create a clear path forward.
AI Enablement Team • April 2026
The system is split into two projects to keep each focused and within context limits.
Parses your raw data files into a structured JSON package. Upload your workbook, Gong transcripts, SkyPrep exports, and Zendesk CSVs here.
Runs competitive research, plans the narrative, and generates a branded HTML presentation + PPTX. Paste your data summary here.
The Builder's presentation engine (70KB component library, 166KB logo assets, themes, PPTX scripts) fills the context window. Keeping data parsing separate prevents Claude from forgetting presentation instructions mid-build.
| Source | File Type | What It Adds | Priority |
|---|---|---|---|
| Macro Workbook | .xlsm |
Core metrics: scheduling, labor, forecasting, engagement | Required |
| Gong Transcripts | .txt |
Customer voice, priorities, sentiment, quotes | High |
| SkyPrep Training | .csv / .xlsx |
Training completion rates, capability adoption | Medium |
| Zendesk Tickets | .csv |
Support partnership, resolution times, channel analysis | High |
| Manual Metrics | Pasted text | Any metrics typed directly (fallback) | Fallback |
More sources = richer EBR. JSON only is good. JSON + Gong is better. All four file sources together produces the most comprehensive review.
From HotSchedules. Must be .xlsm with all 7 standard tabs. Refresh pivot tables in Excel before uploading.
Recent calls with the customer (last 1-2 quarters). Export as .txt files. Even 2-3 transcripts add significant depth.
Completion reports filtered to the customer account. Save as .csv or .xlsx.
Filter by organization and review period. Include ticket ID, category, channel, status, resolution time, and dates. Export as .csv (UTF-8).
Start a new conversation for each customer EBR.
Drag in the .xlsm, Gong .txt files, SkyPrep exports, and Zendesk .csv all at once.
"Prep EBR package for Torchy's Tacos — QSR segment"
You'll get two files: a .json data package and a .summary.md human-readable brief.
No source files? Skip the Processor entirely. Go straight to the EBR Builder and type or paste your metrics when prompted.
Paste the summary.md content. Upload the JSON if deeper analysis is needed.
The system detects your data, processes it, and starts the workflow.
Data Brief → Research Brief → Deck Plan. Approve or adjust at each step. Nothing is built until you say go.
Open the HTML in your browser. Request changes. When satisfied, say "Approved" to generate the PowerPoint.
The system never builds without your approval. You control the narrative at every stage.
Review what was extracted from your uploads. Confirm accuracy before research begins.
Review competitive and industry intelligence. Add context, flag sensitive information.
Approve or reshape the narrative arc and slide outline. Changes are cheap here, expensive after build.
No special commands needed. Upload the Zendesk CSV alongside your other files. The system detects it automatically, applies inclusion/exclusion rules, separates implementation tickets, and flags email channel violations.
| What You Want | What to Say |
|---|---|
| Start a new EBR | "Build an EBR for [Customer]" / "QBR for [Customer]" |
| Run benchmarking | "Yes, run benchmarking" (when offered) |
| Approve deck plan | "Looks good, proceed" / "Build it" |
| Request changes | "Change slide 3 to focus on forecasting" |
| Approve for PPTX | "Approved" / "Looks good" / "Convert it" / "Go ahead" |
| Skip a phase | "Skip research" (system will flag the tradeoff) |
Checkpoint 3 is where you shape the story. Spend time here. Once you approve, the system builds exactly what you asked for.
"Move forecasting to slide 4 with a bar chart" is faster than "make it better." Specific requests = fewer revision rounds.
The "Their World" slide is what makes the EBR feel personal. Without research, it becomes generic — and that's the old format.
Executives remember the last thing they hear. Tie the recommendation to something the customer already said they care about.
Product spelling: Fourth iQ — lowercase i, uppercase Q. Always.
Questions? Reach out to the AI Enablement Team.