Retail and consumer
Customers in India pay by UPI in seconds and expect delivery within the day, while inventory still updates in batches. We rebuild the data layer on Google Cloud so retail AI can act on live signals, inside controls you can show.
Updated October 19, 2026
Retail AI is only as good as the minute its data was last true.
The assortment, the stores and the site are working. Somewhere a team is trialing personalization, availability promises or a service assistant, and each one is waiting on data that arrives late. For consumer brands selling direct, the same delay sits between the storefront, the marketplaces and the warehouse.
Inventory blind spots
A shopper who gets groceries within the hour expects every retailer's stock count to be true. A number that updates overnight sells what is gone and hides what is there.
What waiting costs: margin lost on both sides of the same number.
Consent
The DPDP Act makes consent and purpose binding for loyalty and marketing data, and the RBI's card-on-file rules mean merchants may no longer store card details. A personalization model built on old data habits fails both.
What waiting costs: a program rebuilt after the consent review, and a season of offers you cannot send.
Online selling
The Consumer Protection (E-Commerce) Rules require clear seller and price disclosures, and the Central Consumer Protection Authority's 2023 guidelines ban dark patterns such as false urgency and drip pricing. An agent that writes offers inherits both.
What waiting costs: a consumer authority notice over a pattern nobody reviewed.
| Regulator or law | What it asks | What it means for an agent |
|---|---|---|
| Consumer Protection (E-Commerce) Rules, 2020 | Disclosure of seller, price and return terms, a grievance officer, and no manipulation of prices. | Offers an agent makes show the same price and terms as the catalog, and each one is logged. |
| Central Consumer Protection Authority, Guidelines for Prevention and Regulation of Dark Patterns, 2023 | A list of banned patterns, including false urgency, basket sneaking and drip pricing. | Personalization and service agents are tested against the listed patterns before release and monitored after it. |
| RBI rules on card-on-file tokenization | Merchants and their partners may not store customer card data; tokens replace it. | Agents never see card numbers. Payment calls go through tokenized APIs behind the gateway. |
| Digital Personal Data Protection Act 2023 and DPDP Rules 2025 | Consent and notice for personal data, purpose limits and breach notice. | Recommendations and offers run only on data the customer agreed to, with the consent reference in the log. |
| CERT-In directions (April 2022) | Cyber incidents reported within six hours; ICT logs kept for 180 days in India. | Agent and API logs stay in India and feed the incident process. |
The data exists. It arrives late, in batches, through point-to-point links nobody can trace, so every AI project starts by waiting for the nightly job. Each new agent that touches customer data or payment pages also lands inside payment card and privacy rules.
If a customer buys the last unit at 2:14 p.m., when do your site, your store and your supplier find out?
2:14 p.m., the last unit
Batch world
The item keeps selling until the nightly job.
Event world
Site, store and supplier update in the same second.
Retailers that move their data from batch to events first can let AI act while the customer is still on the page.
Inventory, order and supplier events on one governed stream, built as event streams on Google Cloud with Pub/Sub.
You hold: real-time stock and order events reaching every channel together.
Vertex AI, BigQuery and dataApigee between the store, the site, suppliers and marketplaces, replacing point-to-point links.
You hold: one API layer with an owner and a policy for each connection.
Apigee and APIsModels on Vertex AI and BigQuery and agents on Gemini Enterprise for replenishment, service and personalization, inside spend and action limits.
You hold: one agent in production on live data with a decision log.
Vertex AI, BigQuery and dataEasy to start
Choose one workflow, name one owner, and give us time with the people who sign it off.
We bring engineers and the gateway design, we work inside your environment under your security rules, and we write the review pack alongside the build.
A good first workflow is availability across channels for one product family. See the fixed-price ways to start.
What the work feels like
One workflow is chosen and one person owns it. The questions your reviewers will ask are written down before any build starts. Your merchandising team sees live availability in the first sprint demo.
Governance goes in before the first agent does: the gateway, the audit trail and the access rules. Your risk team sees them working.
A working demo you can show your own review board, not a status slide.
A decision log that can rebuild any action, and a go-live date that was agreed at the start.
Seen it before
Production readiness check
A short readiness check on a live or planned agent workflow. You get a score and the principles worth another look.
Take the readiness checkNo. We work in the integration and data layer around it.
Spend and action limits at the gateway, and a buyer approval step for anything above them.
With event streams on Google Cloud, built on Pub/Sub, so inventory and order events reach the site, the stores and BigQuery together. If you already run another streaming platform, we connect to it instead of replacing it.
Yes, if the patterns are designed out before release. We test every offer flow against the patterns the 2023 guidelines list, and log what each customer was shown so it can be checked later.
No. Under the RBI's tokenization rules card data stays with the card network and the payment partner. Agents work with tokens through the gateway and never see a card number.
Tell us what you are working on. Our Customer Partnership Team will reply to arrange a conversation.
Get answers to your questions
AI assistant. It can make mistakes, so check anything important with our team.
