Process end-of-day sales photos from registers or POS systems. Extract totals, compare against expected ranges, flag discrepancies, and generate daily summaries for the owner.

Daily Sales Reconciler Skill

End-of-day sales processing from register tape or POS photos. Manager photographs the closing summary, sends it via WhatsApp or any channel. AI extracts totals, calculates payment breakdown, flags discrepancies. Owner gets a clean daily summary every morning.

The Problem

Restaurant and retail owners want to know how the day went — but they're not at the register at closing time. The closing manager either calls with numbers (interrupting the owner's evening), texts a blurry photo (owner squints at it), or nobody reports until the owner asks the next day. Cash discrepancies go unnoticed. High discount rates slip by. The owner is always one day behind.

How It Works

Closing manager photographs the register tape or POS daily summary screen
Sends it via WhatsApp/SMS/Slack with @daily-sales-reconciler
AI extracts all sales data — gross, net, tax, payments, cash count
Flags anomalies — cash short, high discounts, high refunds, unusual ticket average
Owner receives a clean, formatted daily summary

Setup

1. Enable Inbound Messaging

Enable the channel your closing manager uses:

WhatsApp (recommended — photo quality is good)
SMS (works but photo quality varies)
Slack / Teams (if your staff uses these)

2. Contact Profiles

Create profiles for: - Closing managers — inbound enabled, so they can send photos - Owner — receives the daily summary via their preferred channel

3. AI Assistant

Must support image analysis: - Anthropic Claude — recommended, excellent at reading receipts and register tapes - OpenAI GPT-4o — good alternative - Ollama — only if using a vision-capable model (e.g., llava)

4. Install the Skill

Pre-installed at INTOUCH_HOME/skills/daily-sales-reconciler/SKILL.md.

Usage

Closing Manager Sends Photo

@daily-sales-reconciler [attach photo of register tape or POS summary]

Or with a cash count:

@daily-sales-reconciler Cash in drawer: $840.00 [attach photo]

Owner Receives Summary

Via their preferred channel (WhatsApp, email, Slack, SMS):

DAILY SALES SUMMARY — 2026-04-14

  Net Sales:        $4,118.20
  Tax Collected:    $339.75
  TOTAL RECEIPTS:   $4,457.95
  Transactions:     187
  Avg Ticket:       $23.84

  Credit/Debit: $3,562.40 (79.9%)
  Cash: $845.55 (19.0%)

  Cash Over/(Short): ($5.55)

  FLAGS: Cash short $5.55

Flags

Flag	Trigger	Why It Matters
Cash discrepancy	Over/short exceeds $10	Theft, errors, or training issues
High discount rate	Discounts over 5% of gross	Unauthorized discounts or coupon abuse
High refund rate	Refunds over 3% of gross	Return fraud or service problems
Unusual average ticket	Below $10 or above $100	Pricing errors or unusual sales mix
Low transaction count	Significantly below typical	Staffing issue, early close, or system problem

Customization

Edit SKILL.md to change:

Cash discrepancy threshold — change $10 to your tolerance
Discount threshold — change 5% based on your promotion frequency
Refund threshold — change 3% based on your return policy
Average ticket range — adjust $10-$100 to your business's normal range
Additional fields — add tip tracking, labor hours, food cost percentage
Multiple locations — have each location manager prefix with the location name

Multi-Location Setup

For businesses with multiple locations:

Create a contact group per location
Closing manager includes location: @daily-sales-reconciler Store #2 [photo]
Owner gets summaries tagged by location
Create a morning YAML job that queries all previous day's summaries and generates a consolidated multi-location report

Cost

Claude: ~$0.02-0.05 per photo (image analysis)
GPT-4o: ~$0.02-0.05 per photo
30 days/month: $0.60-1.50/month

Pairing with Historical Tracking

For trend analysis, pair with a YAML job:

daily-sales-reconciler skill extracts the data
sql task inserts into a daily_sales table (date, gross, net, tax, transactions, cash_variance)
Weekly: Claude analyzes the last 7 days vs prior 7 days, flags trends
Monthly: generates a report with charts description (sales trend, avg ticket trend, cash variance pattern)

This turns a simple daily check into business intelligence — "Tuesday evenings are your weakest period" or "discounts have increased 40% since the new cashier started."