Read Machbase technical articles.

We have entered an era in which you can ask about data in natural language and let AI draw charts and even build dashboards for you. Machbase Neo’s neo-pkg-llm-chat package is an LLM agent that makes this experience possible. If you have used an earlier version, however, you may have encountered frustrations like these: “I connected a local LLM through Ollama, but it keeps calling the wrong tools…” “Why are the API costs higher than I expected?” “When several people use it to

Machbase is a venture company that has spent the past 13 years developing a new kind of software—a time-series database—from the ground up, gathering every resource we could and working to create a market for it. Recently, while preparing a proposal for an airport corporation, we noticed the time-series database requirements in an earlier specification. The original document named two foreign time-series database products, “InfluxDB” and “TimeScaleDB.” Unsurprisingly, Machbas

The last weakness of 2.x was code written directly by AI. Version 2.0.0 greatly reduced failures by separating features according to intent and automatically correcting invalid input. But one class of problem remained: chart code written directly by AI. Creating a chart requires query code behind the scenes, and that code contains many pitfalls. The AI repeatedly made the same mistakes in handling time values and applying aggregation and interval rules. Lightweight local mode

The government encourages the use of open-source software in public-sector IT projects. Avoid dependence on a particular vendor. Save on licensing fees. Open up technology. These are admirable aims. But admirable aims do not necessarily produce admirable results. A free license does not mean there are no costs. The costs do not disappear: installation, integration, performance validation, security patches, incident response, staff training, version upgrades, data migration, a

How does Machbase follow and put ISA-95 into practice? Let’s step inside a factory. Hundreds of machines run continuously, while their sensors generate temperature, vibration, pressure, and current readings every second. When a production line stops, the person responsible checks PLC logs, examines SCADA displays, and compares them with the production results recorded in ERP to find the cause. Yet one problem keeps coming up in this process. The language spoken by equipment d

Characteristics of Cryptocurrency Trading Data Cryptocurrency trading data has distinctive characteristics as a source for analysis. Exchanges operate around the clock, and popular trading pairs can see dozens or hundreds of trades per second. Collecting trades and order book updates for the top 50 coins by trading value on Binance generates 5,000–7,000 rows per second, or roughly 17 GB per day. Yet the data is easy to obtain. Real-time stock prices usually require a brokerag

This post describes how the new AIoT training method with Machbase Neo is different from the traditional one and what its advantages are....

Inputting DHT11 temperature and humidity sensor and querying data in Machbase

‘Why do we need a time series database?’ by focusing on the limitations of existing relational databases and the economic effects that a ti