DigestEngine

Overview

DigestEngine is an automated reporting and analytics pipeline built to extract high-value business intelligence from everyday restaurant point-of-sale (POS) data. Rather than paying exorbitant fees for closed enterprise API connectors or manually massaging spreadsheets every morning, the engine parses raw report exports, calculates real operational KPIs, and dispatches concise executive digests directly to managers via Telegram and email.

The need

Restaurant managers face a persistent data paradox:

• Every transaction, discount, labor hour, and void is recorded by the POS, yet turning that firehose of raw logs into actionable decisions requires hours of repetitive Excel pivot-table work.
• Proprietary POS vendors frequently charge prohibitive recurring fees or restrict access to their modern APIs.
• Owners often don’t find out about margin erosion, labor overtime spikes, or shift underperformance until the month has already ended.

Product goals

• Bypass API lock-in by reliably parsing standard report exports (CSV, TSV, tabular extracts).
• Clean and normalize messy, inconsistent POS schemas into unified data structures.
• Compute daily operational benchmarks: sales pacing, labor-to-revenue ratios, discount anomalies, and category sales velocity.
• Deliver automated executive briefings directly to leadership’s pocket before morning prep starts.

Constraints & Pragmatic Decisions

• No direct POS API access: Instead of battling closed vendor ecosystems, the system treats standard scheduled or downloaded report files as the primary ingestion interface.
• Flexible deployment: Designed to run seamlessly either in local server environments or containerized on Google Cloud, processing batch jobs in seconds.
• Signal over noise: Avoids 50-page dashboard bloat. The output is a crisp, narrative and bulleted digest highlighting only what demands managerial intervention today.

Architecture & Pipeline

+-----------------------+      +-----------------------+
|  POS Export (Sales)   |      |  POS Export (Labor)   |
+-----------------------+      +-----------------------+
            \                              /
             v                            v
+-------------------------------------------------------------+
|                     Ingestion & Cleaning                    |
|             (Schema Normalization & Sanitization)           |
+-------------------------------------------------------------+
                               |
                               v
+-------------------------------------------------------------+
|                     Calculation Engine                      |
|         - Revenue vs. Labor Ratio    - Sales Pacing         |
|         - Voids & Discount Flags     - Product Mix Velocity |
+-------------------------------------------------------------+
                               |
                               v
+-------------------------------------------------------------+
|                      Synthesis Engine                       |
|           (Markdown / HTML Executive Digest Builder)        |
+-------------------------------------------------------------+
             /                                   \
            v                                     v
+-----------------------+              +----------------------+
|     Telegram Bot      |              |   Executive Email    |
| (Morning Briefing &   |              | (Detailed Breakdown  |
|  Threshold Alerts)    |              |  & Historical Trends)|
+-----------------------+              +----------------------+

Current state

In active production. The engine runs daily batch cycles for client operations, delivering daily briefings and weekly strategic summaries without manual spreadsheet intervention.

Lessons learned

You don’t always need an open API to build modern automation. By meeting legacy business systems at their actual output boundary (exported data), you can deliver immediate intelligence and bypass artificial enterprise friction.