Use Case Intelligence RSS Search use cases Search
IBM Software / SaaS Czech Republic

AddAI

Customer Service Automation

AddAI needed to build banking and telecom assistants that were accurate, grounded in verified data and compliant with customer security requirements, while also reducing the time needed to implement each new assistant.

Read the original story Explore all use cases
50% reduction in unanswered queries
85% accuracy rate for answers generated via AI
reduce the implementation time on any new AI assistant by up to 30%

Reported impact

Business outcome

The new pattern delivered materially better grounded answers, cut unanswered customer questions in half and gave AddAI a repeatable way to launch future assistants faster without manually scripting every response path.

  • Customer Experience
  • Productivity Improvement
  • Speed to Market

Solution

AddAI collaborated with IBM Client Engineering and Technical Sales to build a retrieval-augmented customer-support assistant on IBM watsonx.ai, IBM watsonx Assistant and IBM Watson Discovery. Public bank documents such as contracts, terms and conditions, and price lists were tuned in Discovery so the system could understand domain-specific questions and return evidence passages. IBM watsonx.ai accessed open-source large language models including Llama 2 to contextualize that retrieved content and generate Czech answers for topics such as account access, payments, pricing and account services. Watsonx Assistant then exposed the experience through a conversational front end without requiring every dialog path to be hand-authored. The resulting design gave AddAI a reusable pattern for grounded assistant delivery that preserved verified-data answers, supported strict customer security expectations and reduced the implementation effort for new assistants.

Data flow

Public bank documents are indexed in Watson Discovery, user FAQs are submitted through watsonx Assistant, Discovery retrieves relevant evidence, watsonx.ai and Llama 2 generate a Czech answer through RAG, and the assistant returns a conversational response with grounded references.

Solution architecture

4 components · 3 layers
  1. Ingest
    • IBM Watson Discovery Indexes public bank documents and returns reference passages that ground the generated response.
  2. Compute
    • IBM watsonx.ai Generates contextual answers from retrieved banking content and selected open-source language models.
    • Llama 2 Provides the open-source large language model used by watsonx.ai to enhance and contextualize answers.
  3. Serving
    • IBM watsonx Assistant Delivers the banking assistant through a user-friendly chatbot front end.

Architecture clues

  • AddAI required bank-level security, data exclusivity and the ability to opt out of model training on user conversations.
  • Discovery supplied evidence snippets from bank documents so the assistant could cite the basis of its answer.
  • The pilot focused on a banking customer and produced answers in the Czech language.

Evidence from the source

50% reduction in unanswered customer queries
It responded with 85% accuracy
The assistant utilized the retrieval-augmented generation (RAG) methodology
We use watsonx.ai to access selected open-source large language models like Llama 2
Transparency Evidence & methodology 95/100 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 value95/100
  • 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 specificity20/20
    Measures whether the reported benefit is concrete rather than a generic claim.
  • Quantified impact25/25
    3 metrics describe measurable impact.
  • Deployment scale5/10
    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
    4 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. IBM

    AddAI customer story

    First-party
    https://www.ibm.com/case-studies/addai
    Source type
    Vendor-published customer story
    Published
    May 14, 2024
    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

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
IBM Aug 21, 2025 IBM HR Software / SaaS global IBM HR needed to replace fragmented, multi-channel support and repetitive manual work with a single service experience that could answer policy questions, execute transactions, and route complex cases consistently… 40% reduction in the HR team’s operational costs over the past four years Customer Service Automation AskHR agents · Concur · IBM watsonx Orchestrate IBM Jul 17, 2025 Assisto Technologies Software / SaaS India Assisto needed to analyze high volumes of multilingual, code-mixed customer calls in real time so enterprises could extract sentiment, emotion, and actionable insights without relying on slow manual transcription and… 15%–20% improvement in customer retention and engagement Customer Service Automation IBM Cognos Analytics · IBM Granite · IBM watsonx.ai IBM Feb 25, 2025 NDS Cognitive Labs Software / SaaS Mexico NDS Cognitive Labs needed to help Clip support a fast-growing base of first-time digital-finance users at scale, answering questions about fees, setup, payments, and credit without relying on expensive manual support or… 28% annual increase in chatbot usage by Clip’s users Customer Service AutomationSales Automation IBM Cloud · IBM watsonx · NeuralSeek