Use Case Intelligence RSS Search use cases Search
LangChain Financial Services Global

Klarna

Customer Service Automation

Klarna needed a support system that could absorb escalating customer-service volume across global markets while handling payments, refunds, and escalations with enough precision and traceability to preserve response quality.

Read the original story Explore all use cases
2.5 million conversations to date
2.5 million daily transactions
85 million active users
Automated ~70% of repetitive support tasks

Reported impact

Business outcome

Klarna cut resolution times, automated a large share of repetitive support work, and turned its AI assistant into a more reliable global support channel that frees specialists to focus on higher-value investigations.

  • Cost Reduction
  • Customer Experience
  • Productivity Improvement

Solution

Klarna evolved its flagship AI Assistant into a multi-agent support system built with LangGraph and refined through LangSmith. LangGraph gives the team a controllable architecture for routing requests and handling different support tasks, which helps the assistant branch by scenario instead of forcing every case through one path. Klarna also tailors prompts dynamically so responses stay context-aware while reducing latency and token cost. LangSmith adds the step-by-step traces and LLM-evaluation workflow the team needs to test critical support cases, understand why failures happen, and iterate on prompts before issues scale in production. Those learnings also shaped LangSmith features such as meta-prompting, which lets teams propose prompt changes and immediately observe how the optimized prompt affects answer quality. The result is a measured agent workflow, not just a generic chatbot.

Data flow

A support request enters the AI Assistant, LangGraph routes it through the appropriate task path with scenario-specific prompts, the assistant resolves or escalates the case, and LangSmith records step-by-step behavior and evaluation results so the prompts and agent design can be refined.

Solution architecture

3 components · 3 layers
  1. Serving
    • AI Assistant Handles customer payments, refunds, and escalation-related support tasks for Klarna users.
  2. Orchestration
    • LangGraph Routes requests through a controllable multi-agent architecture with task-specific behaviors.
  3. Governance
    • LangSmith Provides traces, evaluations, and prompt-iteration tooling for production hardening.

Architecture clues

  • Context-aware prompt tailoring is used to cut both token cost and latency.
  • Klarna uses a multi-agent architecture rather than a single monolithic prompt.
  • LangSmith traces were important for seeing step-by-step how the assistant behaved.

Evidence from the source

Automated ~70% of repetitive support tasks
Reduced average customer query resolution time by 80%
With 2.5 million conversations to date
seeing step-by-step how their AI assistant behaved
the work equivalent of 700 full-time staff
Transparency Evidence & methodology 96/100 reference value · 89% confidence · 1 source

Review how this reference was scored and trace each normalized claim back to its documented source.

01

Scoring methodology

How scores are calculated

Customer Reference Value measures usefulness as a customer reference. Evidence Confidence measures the quality and traceability of available evidence, not independent verification.

Customer reference value96/100
  • Problem clarity20/20
    Measures whether the customer's operational problem is specific enough to compare.
  • Implementation depth25/25
    Rewards named components, data flow, integrations, and architecture detail.
  • Outcome specificity16/20
    Measures whether the reported benefit is concrete rather than a generic claim.
  • Quantified impact25/25
    2 metrics describe measurable impact.
  • Deployment scale10/10
    Rewards evidence of a live or scaled deployment and meaningful integration scope.
Evidence confidence89%
  • Source authority20/25
    The main evidence is an official vendor-published customer story.
  • Claim traceability21/25
    10 of 12 key fields link to a named source.
  • Direct evidence20/20
    5 direct source quotes retained.
  • Date provenance10/10
    The publication date comes from the source page or publisher.
  • Source transparency8/10
    1 source reference document publisher, URL, authority, retrieval, and supported fields.
  • Implementation specificity10/10
    Rewards source-backed components, data flow, quantified impact, and a specific outcome.
02

Traceability

Sources & provenance

Field coverage shows which facts each source supports. Dates are shown only when documented by the source.

  1. LangChain

    Klarna customer story

    First-party
    https://blog.langchain.dev/customers-klarna/
    Source type
    Vendor-published customer story
    Published
    Feb 12, 2025
    Supports
    • Customer
    • Business problem
    • Solution
    • Products
    • Solution components
    • Data flow
    • Integrations
    • technical area
    • use case category
    • Business outcome
    • outcome category
    • Metrics
    • Architecture
    • evidence quotes
    • Publication date

Date evidence

  • Published on source page Feb 12, 2025 · LangChain · First-party
    Open date source
Not separately sourced Industry, Region, Company size. Treat these normalized attributes as lower-confidence until a dedicated source is added.

Related

Similar use cases

See all like this
OpenAI Oct 16, 2026 Itaú Financial Services Latin America Itaú needed to make financial guidance simpler and more personalized for millions of customers while also scaling document-heavy internal workflows, all under the governance, observability, and economic controls… 100% of the customer base is the goal by the end of 2026 Customer Service AutomationDocument Processing Customer DNA · IARA · OpenAI models Google Cloud Modified · Oct 9, 2026 Nova Gestões Financial Services Brazil Nova Gestões used Gemini to eliminate manual data entry and boost agent productivity by 30% across 100 million monthly calls. 30% productivity increase across 1,700 agents Customer Service Automation Gemini models · Gemini platform Google Cloud Archived · Sep 17, 2026 Home Credit Bank Financial Services Kazakhstan Home Credit Bank needed a safer enterprise AI environment that non-technical teams could use to build banking agents quickly without creating data-leak risk or relying on a long custom-development cycle for each new use… Achieved 100% migration to a secure AI environment Customer Service AutomationEnterprise Knowledge Search Agent Designer · Gemini Enterprise