Customer Service Automation · Enterprise Knowledge Search · Sales Automation
Rakuten wanted to augment both business clients and employees with AI products for market analysis, support, and knowledge work while keeping control over prompt quality, model experimentation, and enterprise data separation across a sprawling multi-business environment.
Reported impact
Rakuten stood up both customer-facing and employee-facing AI products quickly, created a repeatable enterprise prompt-and-eval process, and set the platform up to improve productivity across merchants, hotels, retailers, and internal teams.
Rakuten built a portfolio of AI products on LangChain and LangSmith rather than a single assistant. Rakuten AI for Business serves clients with AI Analyst for market intelligence and chart-backed research, AI Agent for self-serve marketplace support, and AI Librarian for real-time answers over merchant documentation. Internally, the team used LangChain’s OpenGPTs package to stand up an employee platform where teams can build chatbots over internal knowledge. LangChain supplied the reusable chains, agents, and interoperability Rakuten needed to iterate quickly without locking into one model vendor. As usage scaled, LangSmith Hub became the mechanism for distributing the best prompts across teams, while LangSmith Testing and Eval ran custom metrics across models and cognitive architectures inside Rakuten’s own environment with separate development and production access controls.
Business clients or employees interact with the relevant Rakuten AI product, LangChain chains and agents pull the right prompts and document context, and LangSmith Hub plus Testing and Eval feed better prompts and experiment results back into the stack while keeping development and production access separated inside Rakuten’s environment.
By using LangSmith Hub, we could distribute the best prompts
By using LangSmith Testing and Eval with our custom evaluation metrics, we can run experiments on multiple approaches
It only took three engineers one week to get the initial platform up and running.
improve productivity by 20%
intend to roll the product to 32k employees
Review how this reference was scored and trace each normalized claim back to its documented source.
Scoring methodology
Customer Reference Value measures usefulness as a customer reference. Evidence Confidence measures the quality and traceability of available evidence, not independent verification.
Traceability
Field coverage shows which facts each source supports. Dates are shown only when documented by the source.
Related
In-house Sample Apr 8, 2025 Bank of America Bank of America needed to improve employee productivity and client service at scale across a global workforce of 213,000 people, while reducing routine support demand, accelerating employee access to information,… 20 million clients now actively using the virtual assistant Erica · Erica for Employees · PRIVATE BANK 
OpenAI Sample Modified · Oct 16, 2026 Grupo Boticário Grupo Boticário needed to support a complex multi-brand reseller and franchise ecosystem across many channels while reducing repetitive operational work and preserving the human relationships that drive sales and… 100% of conversations generated in appchat and WhatsApp Enterprise covered since November 2025 OpenAI foundation models · WhatsApp Enterprise · appchat 
Google Cloud Sample Archived · Oct 8, 2026 Smyths Toys Smyths Toys needed to handle growing customer-service volume across email, web, and social channels while giving AI agents access to live order, shipping, refund, and product data without slowing or destabilizing core… Resolved more than 300,000 customer emails BigQuery · Gemini Enterprise · Gemini Enterprise Agent Platform 
