Your Commerce AI Toolkit

From Ad Hoc to Systematic

Over the last five chapters, you have used AI for GST compliance, financial analysis, anomaly detection, and cash flow forecasting. But if every interaction starts from scratch — retyping context, re-explaining your business, re-specifying the format you want — you are wasting time. This chapter helps you build a reusable, organized toolkit of prompts that becomes more valuable the longer you use it.

Think of it like your voucher templates in Tally. A new bookkeeper creates every entry from scratch. An experienced accountant has standard templates for common transactions — they just change the numbers and names. Your AI prompt library is that set of templates.

The Prompt Library: Structure and Organization

Open data/prompt-library-commerce.json for a starter library of 30 prompts organized into six categories. Here is the structure:

Each prompt in the library has four parts:

Example: The "Monthly GST Reconciliation" Prompt

Name: GSTR-2A Reconciliation Report

Template:

I have two datasets for [MONTH] [YEAR]:
1. My purchase register (from Tally export)
2. GSTR-2A downloaded from the GST portal

Match them by invoice number and supplier GSTIN.
Categorize each entry as:
- MATCHED: Both datasets agree on invoice number, amount, and tax
- AMOUNT MISMATCH: Invoice exists in both but amounts differ (show the difference)
- IN BOOKS ONLY: In my register but not in GSTR-2A (ITC at risk)
- IN GSTR-2A ONLY: In GSTR-2A but not in my register (potential missed purchase)

Summary required:
- Total ITC in my books: ₹[AMOUNT]
- Total ITC in GSTR-2A: ₹[AMOUNT]
- ITC at risk (In Books Only): ₹[AMOUNT]
- Action items for each mismatch category

Format as a table sorted by category, then by amount (largest first).

Variables: MONTH, YEAR, ITC amounts

Notes: Run this in the last week of every month, before filing GSTR-3B. The "In Books Only" entries need immediate follow-up with suppliers — their GSTR-1 filing is your ITC claim.

Building Your Core Prompts

Let us build four essential prompts together — the ones you will use most frequently.

Prompt 1: GST Invoice Validator

Validate these [NUMBER] invoices for GSTR-1 filing:

Check each invoice for:
1. GSTIN format: 15 characters, state code matches [STATE]
2. HSN code: Valid code, rate matches HSN master
3. Tax calculation: Taxable value × rate = tax amount
4. Tax type: Inter-state = IGST only, intra-state = CGST + SGST only
5. Place of supply: Matches buyer's GSTIN state code
6. Invoice number: Sequential, no gaps, no duplicates
7. Invoice date: Within the filing period [START_DATE] to [END_DATE]

Output: List of clean invoices (count only) and detailed flags
for each problematic invoice with the specific error and
suggested correction.

Prompt 2: Ratio Analysis Report

Using the financial data for [COMPANY_NAME] for [PERIOD]:

Calculate these ratios:
- Liquidity: Current ratio, Quick ratio
- Profitability: Gross margin, Net margin, ROE, ROCE
- Leverage: Debt-to-equity, Interest coverage
- Efficiency: Inventory turnover, Debtors turnover, Creditors turnover

For each ratio:
1. Show the formula
2. Show the numbers used (in ₹ lakhs)
3. Show the result
4. Compare with [INDUSTRY] industry benchmark
5. One-sentence interpretation

Then write a 150-word management summary highlighting:
- Overall financial health (strong/moderate/weak)
- Top 2 strengths
- Top 2 concerns
- One actionable recommendation

Prompt 3: Anomaly Detection Scan

Scan the transaction ledger for [MONTH] [YEAR] and flag
anomalies using these rules:

1. Round amounts: Payments > ₹[THRESHOLD] that are exact
   multiples of ₹10,000
2. Weekend entries: Any transaction on Sunday or gazetted holiday
3. Duplicate invoices: Same invoice number or same amount to
   same vendor within 30 days
4. Threshold splitting: 3+ transactions to the same party
   within 7 days, each below ₹[LIMIT], totaling above ₹[LIMIT]
5. Spike detection: Any amount > 3x the vendor's 6-month average
6. Vague narrations: Entries above ₹10,000 with blank, single-word,
   or generic narrations ("misc", "adjustment", "expense")

For each flag, show:
- Rule triggered (with severity: Critical/High/Medium/Low)
- Transaction details (date, ledger, amount, narration)
- Why it is suspicious
- Recommended verification action

Sort by severity, then by amount (largest first).

Prompt 4: Cash Flow Forecast

Using the last [N] months of cash flow data for
[BUSINESS_NAME], forecast operating cash flow for the next
[FORECAST_MONTHS] months.

Consider:
1. Average monthly pattern from historical data
2. Seasonal factors: [LIST_RELEVANT_SEASONS — e.g., Diwali spike
   in Oct, wedding season Nov-Feb, monsoon slowdown Jun-Aug]
3. Known upcoming changes: [e.g., "new customer onboarding
   expected to add ₹3 lakh/month from March"]
4. Current receivables: ₹[AMOUNT] outstanding, [DAYS] average
   collection period

Output three scenarios:
- Optimistic (+[X]% revenue growth, collections improve by [Y] days)
- Base (same as last year's pattern)
- Pessimistic (-[X]% revenue, [Z]-day delay in collections)

For each scenario, show monthly: inflows, outflows, net cash,
cumulative position. Flag any month where cash goes below
₹[MINIMUM_BALANCE].

Tax Planning Prompts

Open data/tax-planning-scenarios.json — this file contains 8 case studies covering the most common tax planning situations for Indian businesses and professionals.

Section 44AD Presumptive Taxation

My client is a [BUSINESS_TYPE] proprietor with gross receipts
of ₹[AMOUNT] for FY [YEAR]. They are eligible for Section 44AD
presumptive taxation.

Compare two options:
Option 1: Section 44AD — presumed income at 8% of gross receipts
(6% for digital receipts)
Option 2: Regular books — actual profit of ₹[ACTUAL_PROFIT]

For each option, calculate:
1. Taxable income (after standard deduction if applicable)
2. Tax liability under the applicable slab or new tax regime
3. Advance tax requirements (44AD has different rules)
4. Compliance burden (what books/records are required)

Recommend which option saves more tax and reduces compliance.
Note any risks — e.g., once you opt out of 44AD, you cannot
re-enter for 5 years.

Advance Tax Calculator

Calculate the advance tax schedule for [CLIENT_NAME] for
FY [YEAR]:

Income sources:
- Business income: ₹[AMOUNT] (estimated annual)
- Interest income: ₹[AMOUNT]
- Rental income: ₹[AMOUNT] (if applicable)
- Capital gains: ₹[AMOUNT] (if applicable)

Deductions claimed: [LIST — e.g., 80C: ₹1.5 lakh, 80D: ₹25,000]

Calculate:
1. Total taxable income
2. Tax liability (old regime vs new regime — recommend the better one)
3. Advance tax due dates and amounts:
   - 15 June: 15% of tax
   - 15 September: 45% cumulative
   - 15 December: 75% cumulative
   - 15 March: 100%
4. Interest under 234B and 234C if any installment is missed

Versioning Your Prompts

Your first version of a prompt is rarely the best. Over time, you will discover improvements — adding "use Ind AS format" produces better financial reports, or specifying "sort by amount descending" makes the anomaly report more actionable.

Keep a simple version log:

Testing Before Using

Before relying on any AI output for client work, bank submissions, or tax filings, run this checklist:

Building Your Toolkit Over Time

Your prompt library should grow as your practice grows:

The most valuable thing you can do is share your library with colleagues. A GST validation prompt refined for a trading firm can be adapted for a manufacturing firm in 2 minutes. A tax planning prompt that covers Section 44AD can be extended to cover 44ADA (for professionals) with one additional variable. When a CA firm shares a prompt library, every team member benefits — and the firm delivers more consistent, higher-quality work.

Your Complete Toolkit Checklist

By the end of this course, your toolkit should include:

Start with the prompts in data/prompt-library-commerce.json, customize them for your specific business type and client base, and build from there. Within a quarter, you will have a toolkit that saves you hours every week — hours you can spend on the advisory, strategy, and relationship work that commands premium fees and builds a lasting practice.

Key Takeaways