CRED AI Implementation for Premium Customer Experience

CRED integrates OpenAI models to scale premium concierge services

CRED, an India-based members-only club for creditworthy individuals, has transitioned to an AI-first organization to maintain high service quality for its 15 million monthly users. By leveraging OpenAI models—including GPT-4.0, GPT-5, and o3—the company has implemented a suite of AI tools designed to move from transactional customer support to empathetic, concierge-like conversations.

AI-powered customer and internal support tools

CRED has deployed three primary AI tools to optimize the user experience and internal operations:

Cleo: AI Conversational Companion

Cleo is a customer-facing AI companion powered by OpenAI models. It is designed to handle three specific categories of queries:

  • Informational: General questions such as "What is CRED Cash?"
  • Contextual: User-specific eligibility questions, such as "Am I eligible for CRED Cash?"
  • Transactional: Process-oriented queries regarding refunds and payment methods.

Cleo operates by diagnosing the issue, classifying user intent, mapping the query to the correct Standard Operating Procedure (SOP), and generating a contextual response.

Thea and Stark: Internal Operational Tools

To support its employees, CRED developed two internal AI utilities:

  • Thea: A tool for support agents that summarizes multi-format conversations—including text, voice, and Hinglish—and suggests appropriate next steps.
  • Stark: A tool for operations teams that enables the creation or updating of SOPs in minutes rather than days.

Quantifiable impact on customer satisfaction and efficiency

Since the launch of these AI tools, CRED has reported significant improvements in key performance indicators over a three-month period:

  • CSAT Scores: There has been a 14 percentage point improvement in CSAT scores.
  • Resolution Accuracy: Cleo has achieved a 98% resolution accuracy rate.
  • Multi-intent Conversations: Successful resolution of multi-intent conversations has increased by 18%.
  • User Retention: Session drop-offs have decreased by 31%.
  • Handling Time: Average handling times have decreased across Cleo, Thea, and Stark.

Future AI roadmap and organizational goals

CRED is expanding Cleo across all business lines and developing systems to detect "data dead-ends"—queries that cannot be answered—to feed them back into the knowledge base for real-time SOP improvement. The company's broader objective is to make every team member across engineering, QA, infrastructure, and compliance 10x more efficient.

"The question we ask is, how do we make every single member in every function be 10X? For us, AI has become a huge unlock in that journey."

— Swamy Seetharaman, CRED

Sources