Skip to main content

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.

The features on this page are available on Dedicated plans. They extend the core platform for organizations running catalogs at scale, with requirements around custom models, real-time predictions, human feedback loops, and bespoke integrations.

Taxonomy generation & dynamic ontology

Generate sophisticated multi-level taxonomies from your data and keep them evolving as your catalog grows.
  • Multi-level hierarchy — 3–8 levels deep based on data complexity, auto-detected.
  • Semantic relationship mapping — relationships between nodes are explicit and queryable.
  • Dynamic ontology evolution — as new data patterns emerge, Claro proposes taxonomy extensions for review.
  • Cross-domain merging — consolidate taxonomies from multiple catalogs or data sources into a unified structure.
  • Industry-standard templates — GS1, Google Product, UNSPSC as starting points.
  • Validation — automated consistency checking flags contradictions, orphaned nodes, and coverage gaps.
For the standard (non-Dedicated) version, see Taxonomy and the Generate Taxonomy operation.

Entity deduplication & resolution

Enterprise-grade entity matching that goes beyond string comparison.
  • Fuzzy string matching — configurable similarity thresholds per field.
  • Phonetic matching — handles name variations and misspellings across locales.
  • ML-powered semantic similarity — embedding-based matching for product names, descriptions, and addresses.
  • Cross-field correlation — match entities across different attribute combinations, not just a single key.
  • Probabilistic record linkage — scored match proposals with explicit confidence.
  • Negative-example learning — rejected pairs are remembered and suppress similar proposals.
  • Custom matching rules — domain-specific business logic (e.g. match on GTIN first, fall back to brand + MPN).
For the standard version, see Similarity & Duplicate and Find Duplicates.

Near real-time predictions (< 200 ms)

Ultra-low-latency prediction endpoints for production environments where speed is critical.
  • Dedicated GPU clusters with model caching — no cold start.
  • Edge deployment — geographic optimization for latency-sensitive use cases.
  • Auto-scaling — matches request volume without provisioning.
  • SLA guarantees — 99.9% uptime with real-time monitoring and alerting.
  • Load balancing across multiple regions.
Use cases: Live product recommendations, real-time content moderation, instant classification at point of entry, dynamic pricing feedback.

Human-in-the-loop feedback system

Systematic expert review that makes models continuously better over time.
  • Review interface — domain experts approve, correct, or flag AI outputs with an annotation tool.
  • Confidence calibration — expert agreement is tracked and used to refine scoring algorithms.
  • Automated retraining — daily (or configurable: hourly / weekly) model updates incorporating all accumulated feedback.
  • Performance tracking — dashboards showing improvement over time per operation and per attribute.
  • Multi-expert consensus — track agreement across reviewers for contested cases.
  • A/B testing framework — compare model versions before promoting to production.
  • Expert validation — optional gating of model updates behind expert sign-off before deployment.
For standard review workflows, see Notifications.

Advanced API integration hub

Connect Claro to proprietary datasets and specialized external services.
  • Proprietary dataset access — direct connections to internal databases, data lakes, and enterprise systems.
  • Domain-specific APIs — industry databases, patent systems, regulatory filings, ERP and PIM solutions.
  • Auth management — OAuth, API keys, certificate-based, and custom schemes.
  • Data transformation pipelines — ETL for complex format conversions before values land in the catalog.
  • Intelligent caching — responses are cached to reduce cost and improve latency on repeated lookups.
  • Rate-limit management — quota handling across multiple providers, with backoff and retry.
For standard inbound and outbound connectors, see Data Import & Ingestion and Distribute.

Custom analytics

A BI suite surfacing catalog health, operation performance, and team activity across the workspace.
  • Interactive dashboards — configurable per team and per catalog.
  • Real-time monitoring of operation run metrics.
  • Data quality trends over time.
  • Credit consumption and efficiency breakdowns.
  • Export to BigQuery or BI tools via Webhooks.

Self-hosting

Run Claro in your own infrastructure:
  • Kubernetes Helm chart — deploy to any cluster.
  • Managed VPC — Claro sets up a dedicated environment in your cloud account.
  • Air-gapped — fully isolated deployment with no external calls.
  • On-premises — bare-metal or private cloud.
Contact your success team for deployment architecture and sizing guidance.

Getting access

All features on this page require a Dedicated plan. Talk to us at hello@getclaro.ai or schedule a call from the dashboard. We’ll scope the right configuration for your catalog volume and use case.