Gencore AI for Google Cloud Storage

Gencore AI helps organizations rapidly and safely build enterprise-grade generative AI (GenAI) systems, copilots, and AI agents. It streamlines GenAI adoption by enabling the creation of AI pipelines that combine structured and unstructured data from proprietary enterprise sources across hundreds of systems and applications.

You can leverage any foundation model in Google Cloud, such as Vertex AI, Gemini, or PaLM 2, as well as third-party models like Anthropic Claude, Meta Llama 2, and Mistral.

Core Capabilities:

1. Create Safe Enterprise AI Copilots: Build AI copilots and knowledge systems by merging data from multiple sources in minutes, with automated enterprise controls, AI usage monitoring, and full provenance tracking.

2. Securely Sync Data to Vector Databases: Ingest and synchronize enterprise data at scale, generating custom embeddings with metadata for vector databases to prepare data for LLM usage.

3. Curate and Clean Data for Model Training: Assemble, cleanse, and sanitize high-quality datasets for AI model training and fine-tuning.

4. Protect AI Interactions: Utilize the conversation-aware LLM Firewall to safeguard user prompts, responses, and data retrievals.

Key Features:

1. Enterprise-Wide Data Connectivity: Safely ingest data through hundreds of native connectors, supporting AI applications that leverage both structured and unstructured data across SaaS, on-premise, public, and cloud environments.

2. Inline AI Pipeline Controls: Ensure secure AI operations with data sanitization before model consumption, LLM firewalls for policy-compliant interactions, and continuous compliance monitoring (e.g., NIST AI RMF, EU AI Act).

3. Full AI System Provenance: Maintain complete visibility over AI operations, including data and AI usage at the level of individual files, users, and all model endpoints.

Loading Google Cloud Storage Data for GenAI Pipelines:

1. Select Google Cloud Storage as the source.

2. Choose the relevant data system, bucket, or attribute.

3. Define the scope of data to include.

4. Optionally, set filters:

   • Object Prefix: Text or regex prefix match

   • Object Name: Regex-based object name match

   • Object Tags: Key:value tag-based match

   • Object Extension Category or Extensions

   • Object Size

   • Last Modified date

5. Click Save and Continue to proceed to the Data Sanitizer stage.