Overview
Kakao i Dataverse is an Enterprise SaaS platform designed to enable each customer to create and manage their own AI platform where they can train, build, and integrate their own Search Engine and generative AI model.
It integrates Rest APIs with Kakao corp’s LLM KoGPT, 3rd Party engine allowing independent training, storage, and management in a cloud environment.
Research
Main Target User
- B2B contents provider
- Bot & Customer service operator
- Search Engine Operator
Benchmark
- Microsoft : Azure Power app
- AWS : S3
- Google Cloud Platform : Vertex AI Search
Interview
Bot & Search Engine operator
The existing content management, search engine management, bot training, and real-time integration API systems were each separated and not in GUI form, which significantly impacted the workload and efficiency for developers due to the necessity of it.
Contents Provider
Previously, the content upload process required integration through third-party platforms, leading to significant instability in terms of maintenance and limited service provision due to platform performance constraints. It was deemed advantageous to support proprietary storage platforms, various formats, and real-time integration via API standards for convenience.
User research Insights
- Independent authorization system (CP, Bot & Search Management, Enterprise, Individual)
- Proprietary authorization management system
- Real-time connectable Open API
- Easily manageable content & data journey
- Real-time monitoring system, alert functionality for important information
- Easy to connect and learn with 3rd party AI Engine
Service Flow & System Map
When user input in words or a sentence, this platform first finds information using a search engine. Then, it checks if the found info is similar to what it’s already learned. After that, the platform generates new contents based on linked LLM Model. Finally, it serves the contents on connected services.
Features
Search Engine maker - Collection Management
Each customer company can generate own search collection and manage it by using Kakao’s search technology.
User can manage own metadata and entity set for their own search engine database and also can combine vector search.
Search Collection generator
- Each service can generate their own collection
- This collection can connect with the service’s Data Base through rest API.
- User can set up their index rule by combine linked Data Base and Entity.
- This feature can also serve structured search and vector search feature for accuracy of Search result.
Contents Management
Each Customer can manage their contents and connect with their service. And also Every user can connect search engine with contetnts.
Easy Connection
- Can link with the service own DB through rest API and Web Socket.
- Set up batch time as realtime, 5min, 10min …. etc
- Connect with own bot and service directly
Contents Container
- Each service has their own individual contents storage
- Upload Any type of file
- Stream audio and video as realtime
Generative AI Learning & Contents management
Each customer can create their own AI LLM Model based on their uploaded contents. And also can monitor and adjust result for improving their AI Model
AI Contents management
- Can select contents which are already uploaded on Kakao i Dataverse
- Prompt and generate result test
- Can learn model through editing generate result and adding relative prompt
Integrated management
Main dashboard shows integrated information and user can monitor by user data journey and flow through statistics feature
Integrated Dashboard
- Monitoring all changes at once
- Can Access every main feature by one click
- Sending important events and schedule through alert
Data analytics system based on user journey and data flow
- User-centered statistical analysis
- Identifying usage & bounce back rate via user flow
- Analysis user context through usage statistics
- Can gain insights for user-friendly improvements using analysis data
