APAIIF 亞太人工智能產業總會APAIIFAI Knowledge
AI Tools & Applications

Cashfree Relay AI Super Agent Goes Live: India SME Payment Operations Cut from 60 Hours to 45 Minutes Weekly, ₹20 Trillion GMV Loss Problem Gets a Solution

September 6, 20261 Views
Cashfree Relay AI Super Agent Goes Live: India SME Payment Operations Cut from 60 Hours to 45 Minutes Weekly, ₹20 Trillion GMV Loss Problem Gets a Solution
Cashfree
印度AI
支付自動化
AI代理
中小企業

Cashfree Relay AI Super Agent Goes Live: India SME Payment Operations Cut from 60 Hours to 45 Minutes Weekly

India SME's Payment Operations Dilemma

India has over 63 million small and medium-sized businesses (SMBs), which are a vital pillar of the Indian economy. However, these businesses share a common pain point: highly manual payment operations.

According to Cashfree Payments research, Indian SMBs spend an average of 60 hours per week on payment-related operations, including:

  • Manually tracking and retrying failed payments
  • Contacting abandoned cart customers for recovery
  • Confirming cash-on-delivery (COD) orders
  • Managing subscription renewals and expiration reminders
  • Filing payment disputes before deadlines

These repetitive tasks not only consume significant human resources but also result in substantial GMV (Gross Merchandise Value) losses due to delays and omissions in manual operations. Cashfree estimates that the Indian market loses approximately ₹20 trillion (about $24 billion) in GMV annually as a result.

Relay: From Beta to General Availability

In May 2026, Cashfree Payments launched a beta version of Relay, open to select merchants. After four months of testing and optimization, Relay officially opened to all Cashfree merchants in September 2026, becoming a fully available AI super agent platform.

Relay's core positioning is: not just flagging problems, but directly executing solutions.

Traditional payment automation tools typically only identify problems (such as failed payments) and then notify humans to handle them. Relay directly executes corresponding actions using merchant transaction data, without human intervention.

Relay's Core Functions: End-to-End Payment Operations Automation

1. Automated Failed Payment Retry When a payment fails, Relay automatically selects the optimal retry timing and method based on the failure reason (insufficient balance, network issues, expired card, etc.), rather than simply retrying immediately.

2. Abandoned Cart Recovery When customers abandon their shopping carts during checkout, Relay automatically analyzes customer behavior and sends personalized recovery messages through appropriate channels (SMS, WhatsApp, email) at the optimal moment.

3. Cash-on-Delivery (COD) Confirmation COD is an important payment method in Indian e-commerce but also has the highest return rate. Relay automatically contacts COD order customers for confirmation, reducing invalid delivery rates.

4. Subscription Management Automatically handles subscription renewal reminders, expiration notifications, and failed charge retries, ensuring subscription revenue stability.

5. Dispute Filing Automatically identifies payment disputes that need to be filed before deadlines and completes the filing process, avoiding revenue losses from late filings.

Security Design: Merchant Data Never Leaves Cashfree Infrastructure

With AI agents automatically executing financial operations, security is one of merchants' top concerns. Relay adopts strict security measures by design:

  • Data isolation: Merchant transaction data is processed entirely on Cashfree's internal infrastructure, not shared with any external AI providers
  • Human approval at critical junctures: Before critical operations involving fund movement or customer contact, Relay requires explicit merchant approval
  • Operational transparency: All automatically executed operations have complete audit logs that merchants can view and audit at any time

Configuration: Natural Language-Driven Agent Setup

Another highlight of Relay is its natural language configuration capability. Merchants can describe the tasks they want the agent to perform in plain language (including Hindi and other Indian languages), without writing complex rules or code.

For example, a merchant can say: "When a payment fails, retry once after 2 hours; if it still fails, try again after 24 hours, and notify the customer via WhatsApp after the second failure." Relay automatically converts this description into an executable workflow.

Pricing Strategy: Free to Start, Outcome-Based Pricing

Currently, Relay is offered free of charge to all Cashfree merchants. Cashfree plans to transition to an outcome-based pricing model as adoption increases — charging based on actual business results Relay delivers (such as successfully recovered transaction amounts, number of disputes resolved).

This pricing strategy reflects Cashfree's confidence in Relay's effectiveness and lowers the adoption barrier for merchants.

Asia-Pacific Perspective: The Rise of India's AI Agent Ecosystem

Relay's launch is a microcosm of the rapid rise of India's AI agent ecosystem. India has a unique payment environment:

  • UPI (Unified Payments Interface): Monthly transaction volume exceeds 20 billion, making it one of the world's most active instant payment systems
  • High proportion of COD: Despite rapid growth in digital payments, COD still accounts for a significant portion of Indian e-commerce orders
  • SME digital transformation: India's Digital India initiative has driven digitalization of many SMBs, but these businesses often lack technical talent

Relay's emergence provides these digitally transforming Indian SMBs with a low-barrier, high-efficiency AI agent solution.

Conclusion: AI Agents Reshaping India's Payment Operations

The official launch of Cashfree Relay marks an important milestone for AI agents in India's fintech sector. By compressing 60 hours of weekly manual work to 45 minutes, Relay not only improves SMB operational efficiency but also allows these businesses to focus their limited human resources on more valuable business development.

For Asia-Pacific fintech practitioners, Relay's successful model provides an important reference: AI agent applications in payment operations don't require complex technical infrastructure — they just need a deep understanding of local market pain points and needs.

FAQ

Related Articles

Lumafield Quality Agent Officially Launched: World's First AI Quality Agent Powered by Industrial X-Ray CT Foundation Model, Addressing Manufacturing Quality Costs at 5% of Revenue
AI Tools & Applications

Lumafield Quality Agent Officially Launched: World's First AI Quality Agent Powered by Industrial X-Ray CT Foundation Model, Addressing Manufacturing Quality Costs at 5% of Revenue

Lumafield launched Quality Agent on September 3, 2026 — the world's first AI quality agent powered by a foundation model trained on industrial X-ray CT data. The system automatically detects 'unknown unknowns,' performs root-cause analysis, and integrates X-ray, machine vision, environmental sensors, and other data streams. Lumafield research shows 42% of manufacturers spend 5%+ of revenue on quality costs, which Quality Agent aims to dramatically reduce.

Sep 6, 20262
CrowdStrike Falcon Guardian Officially Launched: New Standard for AI Agent Runtime Security, Endpoint Telemetry Traces Complete Agent Execution Graph
AI Tools & Applications

CrowdStrike Falcon Guardian Officially Launched: New Standard for AI Agent Runtime Security, Endpoint Telemetry Traces Complete Agent Execution Graph

CrowdStrike launches Falcon Guardian AI agent runtime security solution at Fal.Con 2026. Uses endpoint telemetry to build complete execution graphs, tracing the full causal chain from user prompts to system actions. Features shadow AI agent discovery, enforceable access controls, and 24/7 managed response services.

Sep 5, 20262
India Sovereign AI Milestone: Gnani Artha Officially Launched, Evon 3.3 Supports 11 Indian Languages with 20% Fewer Tokens Than GPT-5
AI Tools & Applications

India Sovereign AI Milestone: Gnani Artha Officially Launched, Evon 3.3 Supports 11 Indian Languages with 20% Fewer Tokens Than GPT-5

Indian AI startup Gnani.ai launched the Gnani Artha sovereign AI stack on August 28, 2026, presided over by India's Vice President. The stack includes the 30-billion-parameter Evon 3.3 multilingual model (supporting 11 Indian languages) and the Plexus agentic platform. It consumes 20% fewer tokens than GPT-5, with model weights released under Apache 2.0 license, marking a key milestone for India's AI mission.

Sep 4, 202615