DataTalk

General

On this page

General settings — fully expanded sections

The General settings page controls system-wide configuration for the project. Settings are grouped into six collapsible sections: Languages, Alarms, User interventions, AI Engineer, Security, and Delete Project.

Languages

Defines which interface languages are available to users. The project ships with English (en) as the default. Use the Add New Language button to add additional locales — once added, individual users can switch to their preferred language in their own account settings.

Drag the ⣿ handle to reorder languages in the list. Remove an unused language with the × button.

Translating texts

Next to Add New Language there are three more buttons:

Languages section — Add New Language, Export Translations, Import Translations and Autofill translations with AI

ButtonWhat it does
Export TranslationsSaves the texts of the project to a file, for example for an external translator.
Import TranslationsLoads translated texts back into the project.
Autofill translations with AILets the AI translate all texts that are still missing in the additional languages.

Autofill translations with AI uses the AI provider that you set in AI Engineer, so that section has to be enabled and have a valid API key. There is no confirmation question — the translation starts as soon as you click the button, and a window shows the progress:

Autofill translations dialog — progress bar and counters of saved, not translated and skipped texts

  • Missing translations — the number of texts still missing in each language,
  • Translations saved — texts that the AI translated and the project stored,
  • Not translated — texts that the AI did not translate,
  • Skipped (changed meanwhile) — texts that were changed while the translation was running.

When nothing is missing, the window reports All texts are already translated. Newly translated texts are saved in the project immediately, so create a backup in Settings → Backup before you translate a large project, and have the result checked by someone who speaks the language.

Alarms

Disable Alarm Confirmation — When checked, the confirmation step is skipped when an operator acknowledges an alarm. Operators can silence or acknowledge alarms with a single click instead of going through a confirmation dialog.

Storage — The data retention period for alarm history. Entered in the format Xy Xm Xd Xh (years, months, days, hours). The default is 1y 0m 0d 0h (one year). Records older than this limit are automatically purged.

User interventions

Enable — When checked, the system logs operator actions: tag value changes made through the interface are recorded with a timestamp and the identity of the user who made the change.

Storage — The retention period for the intervention log, in the same Xy Xm Xd Xh format. The default is 1y 0m 0d 0h (one year).

Intervention logging and storage

The User interventions storage setting only has an effect when the Enable toggle is checked. If logging is disabled, no intervention records are written and the retention setting is not applied.

Security

Anonymous Autologin — When checked, the system loads without prompting for credentials. Anyone who can reach the server’s URL is automatically signed in. Intended only for physically isolated, fully trusted networks.

Require Two Factor — When checked, two-factor authentication (TOTP) is mandatory for all user accounts. This enforces a site-wide 2FA policy regardless of individual user settings.

Token Expire (hours) — How long an authentication token remains valid after it is issued. The default is 1 hour. Sessions expire after this period and the user must log in again.

User Authentication/Creation Token — A shared secret token that users must supply when self-registering an account. Click Set token to define the token or Clear token to remove it and disable self-registration.

Anonymous Autologin bypasses all authentication

Never enable Anonymous Autologin on a server reachable from the internet or any untrusted network. Every person who can reach the server’s IP address gains full access to the interface and all data without any credentials.

AI Engineer

AI Engineer brings artificial intelligence into the project design. It consists of two independent parts — you can use either of them or both:

  • the built-in AI assistant, a chat panel directly in the Designer,
  • the MCP server, through which you connect an external AI tool of your choice to the project.

AI Engineer section — AI assistant and MCP server settings

Built-in AI assistant

The assistant understands the whole project — assets, tags, connections, graphics, logic and data — and turns your instructions into real project components. Describe what you want to build, change or find, for example “Create a view with a gauge for tag 1.Motor.Speed”.

Setting it up

Enable the assistant

Open Settings → General, expand AI Engineer and tick Enable in the upper part (the assistant).

Choose the AI provider

Select the provider and its model in AI Provider and Model:

ProviderModels offeredDefault
Anthropic Claudeclaude-opus-5-5, claude-sonnet-5, claude-haiku-4-5claude-opus-5-5
OpenAIgpt-5, gpt-5-minigpt-5
Google Geminigemini-2.5-pro, gemini-2.5-flashgemini-2.5-pro
Ollama (runs on your own server)qwen3, gpt-oss:20b, llama3.1qwen3

After you save the settings, the list of models is loaded from the provider. If that is not possible, the default models are shown.

Enter the API key

Paste the API Key of your provider account. Use the eye icon to check what you typed. For Ollama you enter the address of the server instead of a key (by default http://localhost:11434), optionally an access token and the size of the context window (65 536 tokens by default).

Limit the answer length (optional)

Max. Output Tokens limits the length of a single answer of the assistant.

Save

Save the General page. The assistant is active when it is enabled and has an API key (Ollama does not need a key).

Working with the assistant

The assistant opens from the AI Engineer tab on the right edge of the Designer. Write the request into the field at the bottom (Enter sends it, Shift+Enter starts a new line). While the assistant works, the panel lists the individual steps, for example Reading: views or Creating a tag, and then shows the answer. If a step fails, the assistant usually corrects it and tries again.

AI Engineer panel — a question, the step Reading: views and the answer

At the top of the panel you can start a new conversation and open the history of previous ones. Stop interrupts a running answer, and a conversation can be exported to a PDF.

Before you let the assistant change the project
  • The assistant works with the permissions of the logged-in user.
  • Its changes are saved immediately and there is no button to take them back. Before bigger changes create a backup in Settings → Backup.
  • The assistant can make mistakes — check important results.
  • Requests are billed by the provider of your API key. Ollama runs locally and has no per-request cost.

MCP server

The MCP server publishes the project to external AI tools that support the MCP protocol (Claude Desktop, Claude Code, ChatGPT, Gemini, Cursor…). Such a tool can then read and edit the project in the same way as the built-in assistant, but from its own environment — useful when you want to combine the project with other work (files, web, version control) or to make larger batch changes.

Setting it up

Enable the MCP server

In AI Engineer tick Enable in the lower part (the MCP server).

Set the OAuth password

Enter an OAuth Password — at least 5 characters, otherwise the settings cannot be saved. The AI tool asks for this password when it connects for the first time. Save the page.

Add the server to your AI tool

In the AI tool add an MCP server with the address http://<address-of-the-server>/mcpdesigner. For Claude Desktop it looks like this in claude_desktop_config.json:

"datatalk-project-designer": {
  "command": "npx",
  "args": ["-y", "mcp-remote", "http://localhost/mcpdesigner"]
}

mcp-remote needs Node.js. If DataTalk runs on another computer, replace localhost with its address. Outside a local network use HTTPS.

What the AI tool can do

AreaExamples
ConnectionsCreate, edit and delete connections — Modbus, OPC UA, MQTT, EtherNet/IP, databases, REST API, Loxone…
TagsTags, UDT, ƒx templates and formulas, ranges, limit templates.
ViewsSVG graphics, inserting components and symbols, animations, commands, opening views, portrait layout, dark mode, translations of texts.
OtherDashboards, charts, tables and linked tables, scripts and custom components.

What it cannot do

  • It does not see the result. The tool works with the drawing data, not with a picture of the screen — check layouts and colors yourself.
  • It does not read or write live values of tags, manage users and permissions or delete the project. It designs the project; it does not control the running process.
  • Its changes are saved immediately as well — make a backup first.
Another MCP server in the settings

The AI Expert settings of the project contain a separate MCP Server switch for the AI assistant that works with the current data of the running system. It belongs to the AI Module and has nothing to do with AI Engineer.

Which one to use

Built-in assistantMCP server
Best forQuick changes directly in the DesignerLarger batch changes, work together with other tools
You needAn API key of an AI providerAn AI tool that supports MCP and the OAuth password

Delete Project

The Delete Project button permanently removes the entire project, including all configuration, historical data, and user accounts. This action is irreversible and cannot be undone.

Irreversible action

Deleting the project destroys all data. Create a backup in Settings → Backup before proceeding if you need to preserve any part of the project.