Elastic has released version 9.3 of its platform – and it's not just a collection of minor tweaks. Three major features have appeared in the update: built-in workflow automation, chatting with data in natural language, and a builder for creating AI agents. In short, the developers decided that working with data should be easier – so they created tools that don't require constant coding.
Elastic Workflows for Built-In Automation
Automation Inside the Platform
Previously, if you needed to automate something in Elasticsearch – say, process events, react to anomalies, or launch scheduled actions – you had to use external systems or write your own scripts. Now, this can be done right inside the platform.
The new feature is called Elastic Workflows. It is a built-in automation system that allows you to create workflows without needing to deploy additional tools. You can set up triggers – for example, «when a specific type of error appears in the logs» – and link them to actions: send a notification, update an index, or run a script.
This is particularly useful for teams involved in monitoring or working with large volumes of events. Instead of assembling infrastructure from multiple components, you can configure everything in one place.
Natural Language Queries in Elastic 9.3
Ask Data Questions Like You Would a Person
Another innovation is the ability to communicate with data in natural language. This isn't a metaphor: Elastic 9.3 introduces an interface where you can ask a question in ordinary words, and the system itself will figure out how to convert it into a database query.
Let's say you have application logs for the last month. Previously, to find out how many errors occurred in a specific service, you had to construct a query – with filters, time ranges, and aggregations. Now you can simply write: «How many errors were in the payment service over the last week»? – and get an answer.
This approach lowers the barrier to entry for those who don't write queries every day. Analysts, product managers, or those just starting to work with Elasticsearch no longer need to keep syntax in their heads – it's enough to formulate a question.
Of course, this isn't a replacement for full-fledged queries in complex scenarios. But for quick analysis and exploratory tasks – it's a perfectly viable option.
Builder for AI Agents
The third big feature in the update is Agent Builder. This is a tool that simplifies the creation of AI agents. While previously, building such an agent required understanding the architecture, setting up integrations, and writing quite a bit of code, the process has now become much faster.
In this context, an agent is a program that can perform tasks based on user requests. For example, answering documentation questions, searching for information in a knowledge base, or helping with system configuration. Essentially, it's a chatbot with access to data that knows how to do more than just answer with a template – it can analyze context and perform actions.
Agent Builder allows you to assemble such an agent without a deep dive into implementation details. You choose data sources, specify what the agent should have access to, configure behavior – and get a working prototype. Later, it can be refined, but getting started becomes significantly easier.
This is useful for teams that want to embed AI assistants into their products or internal systems but aren't ready to spend weeks developing from scratch.
Key Benefits of Elastic 9.3 New Features
Why All This Together
At first glance, it might seem like these are three different functions that just accidentally ended up in the same update. But if you look wider, there is a common logic here: Elastic is trying to make working with data less technical and more accessible.
Automation removes the need to constantly switch between tools. Natural language chat removes the barrier for those who don't want to learn a query language. The agent builder speeds up the creation of AI applications that previously required serious resources.
All of this – steps towards making the data platform not only powerful but also user-friendly. You don't need to be a data engineer or developer to get value out of it.
What's Next
Elastic 9.3 is already available, and all the described functions can be tried out. Only time will tell how well they work in real-world conditions. Automation might turn out to be insufficiently flexible for complex scenarios, the chat with data will surely make mistakes on non-standard questions, and the agent builder is unlikely to cover every possible case.
But the direction is clear: less code, more focus on the task, lower barrier to entry. If this works, working with data will become accessible to a much larger number of people.