Ottomi Nexus 3.0 - Multimodal AI Data Platform
Functional Module Overview
Ottomi Nexus is an end-to-end data processing platform built on the DataOps philosophy. It is delivered as an all-in-one package and deployed using a containerized architecture, enabling rapid installation, simplified operations, and enterprise-grade scalability.
Its functional architecture is organized into the following major modules:
1. Management Center
| Submodule | Core Capabilities |
|---|---|
| Account Management | User management, organizational unit management, role management |
| Permission Framework | RBAC + ABAC with 6 levels of granularity: System → Project → Data Source → Table → Row → Column |
| Row-Level Access Control | Fine-grained row-level data permission control |
| Log Management | Full operational auditing with tamper-proof logging |
| AI Assistant Configuration | LLM integration settings and API key management |
| System Configuration | Notification channels including in-app messaging, email, and WeCom |
2. Business Planning
| Submodule | Core Capabilities |
|---|---|
| Data Layering Design | ODS raw layer → DWD standardized detail layer → ADS application metric layer |
| Business Domains & Subject Areas | Business domain creation and subject area classification |
| Project Space Management | Project creation, compute source management, member management |
| Dual-Sandbox Architecture | Strong isolation between development sandbox and production sandbox, or optional integrated single-sandbox mode |
| Three Deployment Modes | Large-scale (group-level standard dual-sandbox), medium-scale (hybrid flexible combination), lightweight (all-in-one minimalist mode) |
3. Data Ingestion Engine
| Submodule | Core Capabilities |
|---|---|
| Source Database Management | Registration of heterogeneous data sources; supports 40+ sources including MySQL, PostgreSQL, Oracle, DB2, SQL Server, Dameng, KingbaseES, OceanBase, TiDB, ArgoDB, Greenplum, ClickHouse, Doris, StarRocks, GBase, Hive, and more |
| Database Type Management | Extensible via JDBC drivers |
| Sample Rules & Sample Engine | 5 sample generation strategies: bound sample rules, expression-based calculation, external-table value-domain generation, basic type generation, and original-table-based sampling; 3-layer rule framework (basic / business / special); privacy-preserving transformation with sample data available for computation |
| Resource Exploration | Data source browsing, schema inspection, DDL copy, and data querying |
| Metadata Management | Automatic cataloging and asset publishing/unpublishing |
| AI Auto-Cataloging | AI-assisted automatic cataloging and aggregation of source-side assets |
4. R&D Center · Data Development
Core capability: a visual drag-and-drop ETL canvas combined with an AI-powered conversational assistant that enables modeling through dialogue.
4.1 Development Component Library
9 categories, 95+ components
| Category | Quantity | Representative Components |
|---|---|---|
| Real-Time Input | 7 | Kafka, MySQL CDC, Oracle CDC, SQL Server CDC, MongoDB CDC, PostgreSQL CDC, EventStore |
| Real-Time Output | 3 | Single-table output, StarRocks output, Kafka output |
| Offline Input | 14 | Single table, API, MongoDB, StarRocks, Excel, CSV, XML, Text, S3, JSON, logical table, FTP, SFTP, RabbitMQ |
| Offline Output | 9 | Text, Excel, CSV, XML, JSON, ORC, S3, FTP, SFTP |
| Data Transformation (shared by real-time and offline) | 19 | Outlier detection, unique ID generation, row/column transformation, NULL replacement, data filtering, value replacement, string trimming / case conversion / splitting / concatenation / slicing, field filtering, field name mapping, advanced Java transformation, JsonPath extraction, function computation, encryption/decryption, data masking |
| Offline Scripts | 11 | Script management, SQL, Shell, Python, Flink, MR, FlinkSQL, HQL, DataX, Sqoop, Flink JAR |
| Offline Data Processing | 3 | Aggregation, deduplication, sorting |
| Offline Multi-Table Synchronization | 1 | Batch synchronization of multiple tables |
| Offline Data Fusion | 1 | Table merge |
4.2 Built-in Function Library
84+ built-in functions
| Category | Quantity | Examples |
|---|---|---|
| Numeric Functions | 27 | ABS, CEIL, FLOOR, ROUND, MOD, SQRT, EXP, LN, LOG, POWER, RAND, etc. |
| String Functions | 28 | CONCAT, SUBSTR, TRIM, REPLACE, REGEXP_LIKE, REGEXP_REPLACE, LEFT, RIGHT, LPAD, RPAD, etc. |
| Time Functions | Multiple | Date formatting, date calculation, time difference, etc. |
| System Functions | Multiple | System variables, environment information, etc. |
4.3 AI Canvas Assistant
The Ottomi AI Assistant is built directly into the sidebar of the visual modeling canvas and serves as the intelligent entry point for AI-native workflow construction.
Key Capabilities
-
Natural-language-to-workflow generation
Users simply describe their requirements in natural language. -
LLM-powered intent understanding
The assistant can invoke advanced large language models such as DeepSeek and other integrated cloud or on-premises models to interpret user intent. -
Automatic tool invocation
Based on the user’s request, the system automatically calls built-in platform tools. -
Automatic canvas orchestration
The assistant automatically selects the appropriate canvas components, places them onto the canvas, configures parameters, and wires the workflow logic together. -
User review before execution
Once the workflow is generated, users can open it, review the parameters and variables, confirm that they meet business requirements, and then execute the workflow.
AI-Assisted Scenarios
Ottomi Nexus provides AI-assisted capabilities for:
-
Data Ingestion
Users describe their data collection or synchronization requirements, and the AI assistant automatically builds the required ingestion workflow on the canvas. -
Data Quality Inspection
Users describe quality rules or validation needs, and the AI assistant automatically constructs corresponding quality-check processes and rule logic. -
Data Governance and Data Development
Users describe governance or development requirements, and the AI assistant generates the corresponding visual workflow, including component selection, parameter configuration, and process orchestration.
Supported AI Models
- Supports 18+ AI models
- 14 cloud-based models
- 4 locally deployed models
This makes Ottomi Nexus a true conversational workflow platform, where business requirements can be translated directly into executable visual pipelines.
5. Quality Management Center
Built in accordance with DAMA standards, covering 6 major quality dimensions:
- Completeness
- Consistency
- Accuracy
- Timeliness
- Uniqueness
- Standardization / Compliance
| Rule Category | Quantity | Examples |
|---|---|---|
| Single-Table Structure Checks | 9 | Non-empty table, timestamp fields, complete field comments, primary key integrity, duplicate data, referential integrity, compliant last update time, incremental existence, incremental anomalies |
| Single-Table Field Content Checks | 50+ | Null values, full-width characters, value range, field length, date formats, mobile numbers, national ID cards, passports, bank cards, military officer IDs, email addresses, unified social credit codes, administrative division codes, license plates, blood types, VINs, tax numbers, etc. |
| Single-Table Conditional Checks | Multiple | Combined business rule validation |
| Multi-Table / Whole-Database Structure Checks | Multiple | Cross-table consistency and database-wide standardization checks |
| Multi-Table Dynamic Checks | Multiple | Cross-table dynamic logic validation |
| Real-Time Data Checks | Multiple | Real-time streaming data quality monitoring |
Supported modes:
- Scheduled batch quality inspection
- Real-time streaming quality inspection
- User-defined custom rules
6. Data Asset Management
| Submodule | Core Capabilities |
|---|---|
| Asset Marketplace | A “data supermarket” for browsing, searching, and applying for data assets |
| Data Source Table Assets | Asset cataloging, business classification, lineage tracking, multi-dimensional evaluation |
| Metric System | Atomic metrics, derived metrics, and composite metrics in a three-level indicator framework |
| API Assets | API browsing, application, and approval |
| File Management | Document storage and file upload |
| Intelligent Recognition | OCR, document summarization, and keyword extraction for multimodal data including images, audio, video, and documents |
7. Data Sharing Service Center
| Submodule | Core Capabilities |
|---|---|
| Automatic API Generation | Wizard-based one-click packaging of data tables into RESTful APIs |
| API Marketplace | API publishing, registration, version management, and traffic monitoring |
| Dynamic Masking | Automatic masking during API invocation |
| Approval Workflow | Full lifecycle support for data request → approval → subscription → authorization |
| Interface Marketplace | API online/offline management with customizable approval flows |
8. Data Security and Compliance
| Submodule | Core Capabilities |
|---|---|
| Classification and Grading | Automatic sensitive-data scanning and data classification with S1–S5 levels |
| Encryption | Support for SM2 / SM3 / SM4 Chinese cryptographic algorithms |
| Data Masking | 4 masking algorithms: character masking, encryption (SM4), hash, and character replacement |
| Dual-Sandbox Isolation | Data black box, model white box — production sandbox data remains invisible, the development sandbox uses sample data only, and models can be published to production with one click |
| End-to-End Lineage | Full traceability from source systems to application endpoints |
| Tamper-Proof Auditing | Complete operation logging with hash-based evidence preservation |
| Compliance | Compliant with the Data Security Law and the Personal Information Protection Law |
9. Visual Data Warehouse Modeling
| Submodule | Core Capabilities |
|---|---|
| Kimball Dimensional Modeling | Visual construction of dimension tables and fact tables |
| Drag-and-Drop Cube Design | Multidimensional cubes supporting slice, roll-up, and drill-down |
| Three-Level Metric System | Atomic metrics → derived metrics → composite metrics |
| Database-Agnostic Architecture | Supports any compatible database as the warehouse backend, including MySQL, Oracle, Doris, Greenplum, Hive, and more |
10. BI Analytics and Visualization
| Submodule | Core Capabilities |
|---|---|
| Built-in BI | Integrated based on the open-source DataEase platform |
| Visual Dashboards | Drag-and-drop report creation with no coding required |
| Chart Types | Bar charts, line charts, pie charts, gauges, and large-screen dashboards |
| Self-Service Analytics | Business-friendly analytical interface for self-service exploration |
11. AI Intelligence Center
| Submodule | Core Capabilities |
|---|---|
| LLM Configuration | Integration with public cloud LLMs such as Tongyi Qianwen, ERNIE Bot, and others, as well as privately deployed models |
| AI Agents | Data ingestion agents and data development agents, with editable prompt templates |
| LangChain Orchestration | Collaborative workflows combining multiple tools and LLMs |
| Planned Features | API expansion, MCP (Model Context Protocol) support, and a Skills plugin mechanism |
12. Trusted Data Space
| Submodule | Core Capabilities |
|---|---|
| Zero-Trust Architecture | Connector management and automated deployment |
| Sample Engine | Differential privacy, synthetic data, and format-preserving encryption |
| Space Management | Independent data spaces and compliant cross-space sharing |
| Blockchain Evidence Preservation | Tamper-proof logs plus blockchain-based evidence storage |
13. Task Scheduling Engine
| Submodule | Core Capabilities |
|---|---|
| DolphinScheduler Integration | Distributed task scheduling |
| Scheduling Configuration | Scheduling cycles by second, minute, hour, or day |
| Dependency Orchestration | Upstream/downstream dependency orchestration for complex workflows |
| Monitoring and Alerts | Runtime log monitoring and exception alerting |
14. Operations and Maintenance
| Submodule | Core Capabilities |
|---|---|
| Infrastructure Monitoring | Service and hardware status monitoring |
| Data Backup | Backup of configuration databases and configuration files |
| High Availability | Active-standby architecture with automatic failover |
Product Highlights
AI-Native Conversational Workflow Construction
Ottomi Nexus brings AI directly into the data workflow lifecycle.
Users can simply chat with the Ottomi AI Assistant to describe what they want to accomplish. The assistant then leverages advanced LLMs such as DeepSeek and other supported models to understand intent, automatically invoke platform tools, select the required components on the visual canvas, pass parameters, and build the workflow end to end.
After the workflow is generated, users can open it, review parameters and variables, confirm that the logic matches their requirements, and execute it directly.
This conversational approach enables:
- AI-assisted data ingestion
- AI-assisted data quality inspection
- AI-driven data governance
- AI-driven data development
In short, Ottomi Nexus transforms business requirements into executable workflows through conversation.
Summary
The core product philosophy of Ottomi Nexus can be summarized as follows:
Data Black Box · Model White Box
The dual-sandbox mechanism ensures that data remains secure and controllable, while models remain transparent and auditable.
Modeling Through Conversation
The Ottomi AI Assistant transforms natural language requirements into visual, executable workflows.
All-in-One Delivery
A single Docker Compose command can take the platform from zero to usable in as little as 15 minutes.
Full-Stack Data Processing Coverage
With 9 categories and 95+ components, 84+ built-in functions, and 60+ quality inspection rules across 6 categories, Ottomi Nexus covers the full data processing lifecycle.
Enterprise-Grade Security Baseline
With 6-level permission granularity, 4 masking algorithms, and compliance with Chinese cryptographic standards and data regulations, Ottomi Nexus provides a robust security foundation for enterprise deployment.