AI Service

Enterprise AI services that turn generative AI into secure, practical business outcomes.

Scroll

Saltware AI Brand, FitSapie

Sapie is Saltware's enterprise AI brand, built on proprietary Korean-language models and practical delivery experience.
From large language models and data analytics to workflow automation, we help organizations adopt AI in ways that are secure, scalable, and closely aligned with business needs.
Our technology leadership has been recognized in market evaluations, validating both performance and reliability.

01. Sapie-Bot

Sapie-Bot illustration
Sapie-Bot is an LLM-powered enterprise chatbot designed for 24/7 Q&A, task automation, and employee assistance.
It helps organizations respond faster to customer and internal requests while improving operational productivity and service quality.
AI chatbot illustration

Why Sapie-Bot

  • High-accuracy AI responses

    - Reduces hallucinations through optimized retrieval and response design - Applies enterprise knowledge from multiple internal and external data sources

  • Flexible deployment options

    - Available across cloud, on-premise, or hybrid environments - Can be customized to match each customer's domain, workflow, and service policies

  • Easy integration and expansion

    - Core features are provided through APIs for rapid service extension - Can connect with existing enterprise systems and databases

Service Architecture

AI chatbot service architecture

Advanced RAG

Our advanced RAG architecture improves answer accuracy based on enterprise data while minimizing hallucinations.
With optimized search, ranking, and context orchestration across diverse data sources, Saltware builds RAG services tailored to each customer environment.
Advanced RAG overview

02. Sapie-Reco

Sapie-Reco illustration
Sapie-Reco is Saltware's AI recommendation service that understands user intent and behavior, analyzes product and content data, and generates more relevant recommendation experiences.
It helps businesses improve conversion, engagement, and customer satisfaction with more adaptive recommendation logic.

Why Sapie-Reco

  • Lower development and operating overhead

    - Simplifies the implementation and management of advanced recommendation services using LLM-based intelligence - Improves operational efficiency significantly compared with conventional recommendation approaches

  • Better recommendation performance

    - Enhances recommendation quality across both general and cold-start environments - Increases purchase conversion and user retention when combined with existing recommendation systems

  • Improved customer buying experience

    - Delivers more personalized recommendation flows and shopping journeys - Supports more creative and context-aware category experiences that help customers discover more relevant products

Service Architecture

Sapie-Reco service architecture

03. Sapie-LLM

Sapie-LLM illustration
Sapie-LLM is Saltware's proprietary large language model designed for enterprise environments.
It enables domain-specific AI services while reducing security concerns such as unintended data exposure, helping organizations adopt generative AI more safely and effectively.
Backed by Saltware's model customization capabilities, Sapie-LLM can be fine-tuned and integrated to fit each customer's requirements.

Why Sapie-LLM

  • Domain-specific AI services

    - Supports fine-tuning with customer data to build industry- and business-specific AI services

  • Enterprise-grade security

    - Reduces the risk of data leakage through controlled model operation and secure deployment options

  • Customization and integration

    - Saltware provides tailored model customization services aligned with customer needs - API-based delivery and modular architecture make service integration easier

Performance Benchmark

Sapie-LLM performance benchmark

References

1. Sapie-Bot reference: 24/7 university Q&A chatbot

  • A university operating a KakaoTalk chatbot for student and parent inquiries needed a more scalable way to respond during peak admission periods.
  • Saltware implemented the cloud infrastructure, automated RAG pipeline, model configuration, and operational processes required for a generative AI chatbot service.
  • The service was connected to existing channels and the official website to minimize disruption and accelerate adoption.
  • The chatbot reduced repetitive inquiry workload and improved customer satisfaction by providing around-the-clock responses.

2. Sapie-Reco reference: personalized recommendation service for retail

  • A retail customer wanted to move beyond rule-based recommendation logic and introduce a more adaptive product recommendation experience.
  • Saltware built a data pipeline for product and user information, generated embeddings, constructed a vector database, and implemented customer-facing recommendation UI/UX.
  • The resulting service improved recommendation quality, addressed cold-start challenges, and created more dynamic shopping experiences.

3. Sapie-LLM reference: internal policy chatbot for enterprise operations

  • An IT company needed a way to help employees quickly find internal policy and operational guideline information without overloading support teams.
  • Because the content included sensitive internal information, the company chose a private deployment model using Sapie-LLM.
  • Saltware fine-tuned the model for the customer's domain, built a RAG system with distributed internal documents, and enabled chatbot creation and operations through Sapie Studio.
  • The result was faster internal support, reduced workload for administrative teams, and better organization-wide access to policy knowledge.