Frequently Asked Questions (FAQ)
Quick answers to the most common questions about the DATTA platform — what it is, what it needs to run, who controls what, and what to do when a screen comes up empty. Each answer points to the full guide on the subject, for when you want more than the summary.
What is DATTA?
DATTA is a data and AI platform that runs on your own infrastructure and combines Data Mesh, BI (DATTA BI), intelligent search with chat, automated case screening, a BPM engine and a multi-agent copilot — all in a single product, with no mandatory cloud dependency.
Do I need internet access to use the platform?
Not for the core. All essential components — Neo4j, OpenSearch, Kafka, Trino, the platform cache and the DATTA services — run inside your own environment. The only capability that uses an external API by default is the language model (Gemini). In isolated environments, the administrator can enable local mode with vLLM and open models — the AI features remain available.
What language is the interface in?
The platform interface is in Brazilian Portuguese. This documentation portal is available in Portuguese and English — the language switcher sits at the top of the sidebar and the initial detection uses your browser's language.
What is a Context?
A Context is the knowledge domain you work in (for example: CPC, CDC, tax cases). Each context has its own graph database, search indexes and screening rules — one domain's knowledge does not "leak" into another. The administrator creates contexts through the interface, in — see Manage Contexts.
Why can't I find a dataset in the catalog?
Every dataset must be registered in the Knowledge Catalog, and connections created through the platform trigger automatic cataloging (ACOC). If a dataset does not appear, check in this order:
- Was the connection really created in ?
- Has the scan finished? Check its progress in the Background Executions panel.
- Do you have view permission in the workspace where the dataset was published?
Who controls what I can see and do?
The platform's access control. Roles aggregate named permissions (for example DASHBOARD_WRITE, CONNECTION_CREATE, PLATFORM_VIEW) and every action is validated on the server — hiding a button is never the only barrier. If something appears disabled or returns "access denied", talk to the administrator about your role — see RBAC.
Batch screening won't start. What should I check?
Screening uses the ACTIVE context of the session. Confirm, in this order: (1) the correct context is selected; (2) there are published screening rules for that context; (3) there are cases ingested into the context's database. The step-by-step guide is in Batch Screening.
Which file formats can I upload for ingestion?
PDF (including scanned ones, with OCR), DOCX, HTML, TXT and tabular formats for structured data. The full list, the size limits and the behavior for very large files are in Supported Formats.
The chat "made up" an answer. How do I report it?
Every chat answer displays the retrieved sources. If the answer does not cite sources or cites the wrong ones, use the feedback button on the message itself — this feeds knowledge curation. An answer without a source should not be used as a basis for decisions.
How do I track long-running tasks (ingestion, scans, materializations)?
Through the Background Executions panel, opened from the activity icon at the top of the screen. It shows the queue, progress and result of each task, with status chips such as Executando and Na fila — see Background Executions.
Can I connect external tools (DBeaver, BI, scripts) to the data?
Yes. Neo4j (bolt), Trino (JDBC/HTTPS) and OpenSearch (HTTPS) are exposed with TLS, and the users of each engine are managed through the administration consoles. The connection URIs and ready-made client examples are in the subconsoles of each engine — , and .
Dashboards disappeared from my selector. Is it a bug?
In most cases it is the workspace: dashboards belong to a workspace and the selector shows only the active workspace. Switch the workspace at the top of the shell. If the problem persists after signing out and back in, open a ticket with the administrator.
Does the platform keep a history of my conversations?
Yes, the chat conversation history is persisted and searchable — including semantically. It belongs to your user and respects the context permissions in effect at the time of the query.
Where do I find the operations documentation (runbooks, infrastructure)?
In the Platform & Operations section of this portal, visible only to authenticated users with platform permission (PLATFORM_VIEW). If you operate the environment and do not see the section, request the permission from the administrator.