> ## Documentation Index
> Fetch the complete documentation index at: https://docs.getclaro.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# FAQ

> Frequently asked questions about Claro.

## Catalog & data

**Q: What's the difference between a Catalogue and a dataset?**
A: A Catalogue is persistent — it has a defined schema (Attributes), versioned records, data sources, and a full history with provenance. Operations write back to it on a schedule or in a pipeline. A dataset (from a Research Agent) is a one-off output; it exists until you promote it into a Catalogue.

**Q: Can I have multiple catalogues in one workspace?**
A: Yes. Catalogues can also reference each other — for example, `Product.supplier` → `Supplier`. Cross-catalogue references enable filtering, enrichment, and denormalization across related objects.

**Q: How do I handle large catalogs (millions of records)?**
A: Use database connectors (BigQuery, Postgres, Supabase) or S3 as Data Sources rather than file uploads. Operations run in parallel batches server-side. Contact us for guidance on sharding and scheduling for very large volumes.

**Q: What file formats are supported for import?**
A: CSV and XLSX for file upload. Structured JSON and CSV via HTTPS pull and database connectors. PDFs for Knowledge Bases (Word, PowerPoint, and plain text coming soon).

***

## Operations

**Q: How is the confidence score computed?**
A: Confidence combines four factors: source reliability (40%), content consistency (30%), model certainty (20%), and retrieval quality (10%). See [Confidence Scoring](/confidence_scoring) for details.

**Q: What happens if an operation fails partway through?**
A: Failed rows are retried automatically. Rows that continue to fail after retry consume zero credits. The run shows a partial success count and per-row error details.

**Q: Can I roll back a bulk enrichment that went wrong?**
A: Yes. Every write is reversible from the record's per-field history. Roll-back is per-field and is itself a logged event with your identity as the source.

**Q: What does "chained pipeline" mean?**
A: You can configure operations to trigger each other in sequence — for example, Data Source Mapping → Validate Data → Normalize Data → Bulk Enrichment → Push & Sync. Each stage gates the next; a failed validation stops the pipeline before bad data reaches the catalog.

***

## Plans & billing

**Q: Do unused subscription credits roll over?**
A: No. Monthly subscription credits reset each billing cycle. Top-up pack credits never expire.

**Q: Can I cancel my subscription at any time?**
A: Yes. Cancel from **Settings → Billing**. You retain access until the end of the current billing period.

**Q: Can I supply my own OpenAI API key?**
A: This is planned for a future release. Currently all AI processing runs through Claro's managed infrastructure.

**Q: Is self-hosting available?**
A: Yes, on Dedicated plans. We provide a Kubernetes Helm chart or can set up a managed VPC in your environment. See [Security & Compliance](/security) for details.

***

## Research Agents

**Q: Can I turn a Research Agent dataset into a persistent catalog?**
A: Yes. From any Generated Dataset, choose **Promote to Catalogue**, pick the target catalogue, and map columns to attributes. The records become persistent and subject to all the same operations, history, and review workflows as native catalogues.

**Q: Are Research Agents the same as the old Claro product?**
A: Research Agents are the evolved form of the original dataset-generation features. They've been expanded with new agent types and now sit alongside the persistent Catalogue as one of two complementary surfaces on the platform.

***

## Security & compliance

**Q: Where is my data stored?**
A: EU and US data residency options are available. Region is selected at workspace creation.

**Q: Is Claro SOC 2 certified?**
A: A SOC 2 Type II audit is in progress; expected completion Q4 2025. DPA available on request.

**Q: How do I get support?**
A: Free and Starter plans include community support via our Slack channel. Dedicated plans include a dedicated success team. Email **[hello@getclaro.ai](mailto:hello@getclaro.ai)** for anything urgent.