Contribution · Application — E-commerce & Retail

LLM-Informed Dynamic Pricing for E-commerce

Classical dynamic pricing is a demand-elasticity and reinforcement-learning problem. The LLM layer adds something new: the ability to reason over messy context — competitor promos scraped from the web, a supplier email announcing a price hike, a forecasted demand spike from a cricket match. Deployed as a copilot for pricing managers it can be transformative; deployed autonomously it is a regulatory and reputational minefield.

Application facts

Domain
E-commerce & Retail
Subdomain
Pricing & revenue management
Example stack
LightGBM / CausalML for price elasticity estimation · Claude Sonnet 4.7 for contextual reasoning and copilot UI · Competitor price feeds via Bright Data or DataWeave · Feast feature store for SKU + market features · Rule engine (Drools / custom) for price band enforcement

Data & infrastructure needs

  • Historical transaction data with price and demand signals
  • Competitor price feeds and inventory observations
  • Margin, cost, and MRP data for floor/ceiling constraints
  • Category rules — MRP Act, drug pricing, fuel caps

Risks & considerations

  • Algorithmic collusion — price signals converging across competitors (CCI scrutiny)
  • MRP Act violations in India if listed price exceeds declared MRP
  • Consumer Protection (E-commerce) Rules 2020 prohibit manipulative pricing
  • EU AI Act classifies certain pricing systems as high-risk under profiling rules

Frequently asked questions

Is AI dynamic pricing legal in India?

Yes within limits. The Legal Metrology (Packaged Commodities) Rules and MRP Act cap the maximum listed price. CCI scrutinizes algorithmic collusion. Consumer Protection (E-commerce) Rules 2020 require that personalized pricing not be deceptive. Surge pricing in ride-hailing and travel is separately regulated.

What model is best for dynamic pricing?

The model is not an LLM — it is an econometric elasticity model (LightGBM, bayesian regression, or causal-ML). Claude Sonnet 4.7 acts as a copilot, explaining pricing decisions and handling exceptions, never as the primary decision maker for mass SKUs.

Regulatory considerations for pricing AI?

India: DPDPA (if user-level personalization), MRP Act, CCI anti-collusion rules. EU: AI Act profiling obligations, Digital Markets Act for gatekeepers. US: state-level price-gouging laws during emergencies.

Sources

  1. CCI — market study on e-commerce in India — accessed 2026-04-20
  2. Legal Metrology Act — DoCA — accessed 2026-04-20

Related