Machbase NEO
A data platform for innovative AI.
Three layers that turn physical data into AI value.
WHY PHYSICAL AI NEEDS A NEW DATA FOUNDATION
View different types of data
on the same timeline.
Camera, LiDAR, vibration, current, torque, position and event data differ in format and speed.
But root cause analysis and AI decisions share one reference point:
When did it happen?
VISION
Cameras · LiDAR · Depth sensors
MOTION
IMU · Acceleration · Vibration · Impact
ACTUATOR
Current · Voltage · Torque · Encoder
POSITION
Position · Path · Joint · Trajectory
EVENT
Alarms · Faults · State changes
TIME AXIS
01
High-speed input
Continuous collection
02
Compressed storage
Long-term retention
03
Time lookup
Range Navigation
04
Long-term aggregation
Trends
05
Data retention
Policies and recovery
06
Service delivery
APIs and visualization
Ingestion performance alone is not enough.
Storage efficiency, time-range analysis, the data lifecycle, and service integration all need to be designed together.
A single-executable data platform
cuts software management costs.
ONE PRODUCT · THREE CONNECTED LAYERS
Connect collectors, brokers, databases, API servers, visualization and schedulers within a single product.
PHYSICAL DATA
Sensors
Logs
Video
LiDAR
Robots
01
LAYER 1: DATA
Database Engine
Process raw time-series data and relational business data in a single DBMS, using the right approach for each workload.
Machbase DBMS 8.7 offers high-speed, append-oriented TAG and LOG tables, TRANSACTION tables for relational DML and transactions, VOLATILE tables for real-time state, and LOOKUP tables for reference data.
TAG
Sensors · Equipment
Manage large volumes of measurements and ROLLUP statistics by tag name and time or distance axis.
Columnar · PRIMARY KEY · Persistent
LOG
Events · Logs
Store append-oriented events, packets, and logs with automatically generated arrival timestamps.
Columnar · _arrival_time · Persistent
TRANSACTION
Business· status data
Manage relational business data and aggregated results with row-level DML and explicit transactions.
Row-based · Standard Edition · Persistent
VOLATILE
Real-time state
Quickly query and update shared temporary server state and caches in memory.
In memory · Cleared on restart
LOOKUP
Reference data
Manage small, frequently referenced datasets such as equipment codes and tag metadata.
Row-based · PRIMARY KEY required · Persistent
TRANSACTION TABLE · RELEASED SPEC
Connect raw data and operational state in the same database.
Store high-frequency raw data in TAG and LOG tables, and manage equipment status, inspection results, inventory, and aggregated results in TRANSACTION tables. JOIN time-series data with relational context to understand operational data in business terms.
FULL DML
INSERT · SELECT
UPDATE · DELETE
INDEX & MODEL
PRIMARY KEY · UNIQUE
Secondary indexes · JOIN
TRANSACTION
BEGIN · COMMIT
ROLLBACK
ENTERPRISE DATA
AUTO_INCREMENT · JSON
BACKUP · MOUNT
HIGH-FREQUENCY RAW DATA
TAG · LOG
BUSINESS STATE & CONTEXT
TRANSACTION
REFERENCE & LIVE CACHE
LOOKUP · VOLATILE
High-speed input
Continuously ingest millions of sensor readings, ticks and events per second
Large-scale storage
Storage of over 200 billion sensor data points based on 1TB
ROLLUP
Pre-aggregate trends for seconds, minutes, hours, days, months, and years for quick retrieval
BACKUP / MOUNT
Full, time-range and incremental backups, with recovery that allows immediate queries during operation
STANDARD & NATIVE ACCESS
Access your data using SQL 92 and familiar programming languages.
SQL 92
C#
Phython
Node.js
JDBC
ODBC
Go
02
LAYER 2: SERVICES
Data Service
Connect, transform and visualize collected data, then make it available through APIs, automation and an extensible runtime.
HTTP and MQTT ingestion and queries, external database and message bridges, TQL transformations, dashboards, scheduling and authentication are provided as shared services within the product.
HTTP API
Run SQL through /db/query and ingest JSON, CSV and ILP data through /db/write and /metrics/write.
MQTT Engine
Write data through MQTT v3.1.1/v5 sessions and Append, and receive query results through db/query and db/reply.
Bridge
Connect to SQLite, PostgreSQL, MySQL and MS-SQL for DDL and DML,
and to NATS and external MQTT for messaging.
Timer
Use CRON or @every schedules to automate TQL ingestion, aggregation, transformation, database writes and external transfers.
Security
Use token authentication for HTTP and MQTT, X.509/TLS for MQTT, and manage the key lifecycle through the Web UI.
TQL · TRANSFORMING QUERY LANGUAGE
Combine reading, transformation, and output into a single pipeline.
Without a separate application, process data from SQL, CSV, JSON, HTTP, Bridge and MQTT using filtering, aggregation, FFT and JavaScript, then output to DB, API, CSV, JSON, Chart, HTML or Markdown.
SOURCE
SQL · CSV · JSON
HTTP · Bridge · MQTT
MAP
Filter · Aggregate
FFT · SCRIPT()
SINK
DB · API · Chart
HTML · Markdown
BUILT-IN DASHBOARD
Turn collected data into
operational dashboards instantly.
Manage multiple queries and charts, a shared time range, automatic refresh, variable-based queries and Geomap on a single screen.
Dozens of chart types and dashboards
Timer automation using CRON and @every
Manage task status through the web UI and CLI
03
LAYER 3: AI APPS
AI Application Runtime
Run user applications and AI agents on your data services.
JavaScript-based JSH provides database, HTTP, MQTT, NATS, and OPC-UA modules, so you can develop and run site-specific applications using only the Machbase Neo executable.
BUILD
AI Coding Agent
INSTALL
Built-in / User Apps
RUN
JSH Runtime
ACCESS
Neo Data Service
Machbase Neo 8.5.0 +
Neo directly runs
JavaScript files and app directories.
Files with the .js extension and directories containing index.js are recognized as executable applications. JSH also supports execution independent of the server, a virtual file system and mounting host directories.
JSH
App Structure
.js · index.js directory
Standalone Run
machbase-neo jsh
Built-in Modules
DB · HTTP · MQTT · NATS · OPC-UA
BUILT-IN APP
Data Replication
Selectively replicate data in real time across edge devices, servers and the cloud.
BUILT-IN APP
OPC-UA Client
Select a collection node using Browse, process the data, and save it to a designated table.
BUILT-IN APP
Blackbox
Store and explore video and sensor data on a shared timeline, and track events.
BUILT-IN APP
LLM Chat
Run SQL and TQL using natural language, and generate analytics dashboards and HTML reports.
APP DEMO REPOSITORIES
Industry-specific sample apps you can validate with real code
Explore installation configurations and data flows in Machbase's public repositories.
Robot Arm Simulation
Load robotic arm motion and sensor tag data into Neo and view it in a dashboard.
Railway APU Diagnosis
Visualize railway APU operating data and validate equipment condition diagnostic workflows.
From collection to analysis,
one continuous flow of data.
FROM PLC DATA TO AI ACTION
Connect the complete workflow, from PLC data in OPC UA environments to video and sensor integration, real-time replication, and generative AI analysis.
COLLECT
OPC-UA
Standard industrial data collection
PRESERVE
Database
Long-term storage of raw data and statistics
CONNECT
Replication
Real-time replication between edge devices and servers
ANALYZE
LLM Agent
Natural-language Analytics and Reports
ACT
Operation
Diagnostics · Alerts · Task execution
DOWNLOAD & VERIFY
It's easy.
Run it and check the results.
Download Machbase Neo to try data ingestion, SQL queries, TQL transformations, charts and dashboards yourself.
NEO
01
Download
Download the package for your operating system.
02
Run
Start the service with a single executable.
03
Verify
Verify the overall flow with actual data.
