Announcing Support for All AI Models in Amazon Bedrock in Securiti’s Gencore AI

We are delighted to announce Securiti’s Gencore AI integration with Amazon Bedrock, which enables enterprises to safely use their unstructured and structured data from hundreds of systems through Amazon Bedrock's rich library of AI models, powering innovations like enterprise copilots, knowledge systems, and business process automations. Gencore AI is now available through AWS Marketplace.

.

Amazon Bedrock is a fully managed service that offers a choice of high-performing foundation models (FMs) for building generative AI applications. Gencore AI supports the following AI models in Amazon Bedrock:

• Amazon Titan Text G1 - Express
• Amazon Titan Text G1 - Lite
• Anthropic Claude 3 Sonnet
• Anthropic Claude Instant 1.2
• Anthropic Claude 2
• Anthropic Claude 2.1
• Meta Llama 2 Chat 13B
• Meta Llama 2 Chat 70B
• Meta Llama 3 70B Instruct
• Meta Llama 3 8B Instruct
• Mistral 7B Instruct
• Mistral Large
• Mixtral-8x7B Instruct

Through this integration, enterprises can leverage their proprietary data with any AI models in Amazon Bedrock while maintaining comprehensive security, governance, and compliance. This is achieved through secure data synchronization and sanitization at the data ingestion level, and strict enforcement of enterprise policies, controls, and entitlements at the AI consumption level.

To enable rapid genAI innovation with proprietary data at scale, four key considerations have to be accounted for.

1. Easy to Build and Operate Safe AI Systems:

A typical enterprise has dozens or hundreds of genAI use cases to be implemented and operated.  To implement, deploy and operate these genAI projects at scale, software tooling is needed that makes it easy to utilize unstructured and structured data in diverse systems with genAI models.

Gencore AI makes it easy to connect to unstructured or structured data in diverse data systems and utilize any genAI model in Amazon Bedrock.

AI teams can configure and operationalize these systems in minutes.

2. Embedded Data Security & Governance in AI Systems:

Security, governance and safe use of proprietary data is the top need and baseline requirement for CIOs and CISOs for genAI projects. It's a key concern in moving from proof of concepts to production-ready, enterprise-grade systems. This is not surprising, because at the end of the day, AI is fueled by data, and it is not possible to build safe AI systems without diligent oversight of that data. In a recent survey of senior leaders by Deloitte, top concerns include using sensitive data in models, managing data security privacy related issues and complying with regulations.

Other surveys reflect similar concerns. What is particularly challenging is that most of the data used for building genAI systems will be from unstructured data systems, as it represents 80% to 90% of an organization's data estate. While an ecosystem in the industry has evolved for managing structured and semi-structured data over the years, governing unstructured data is generally uncharted territory.

Gencore AI aligns with OWASP Top 10 for LLMs to help embed data security and governance at every important stage of the AI System, from data ingestion to AI consumption layers.

Gencore AI automatically sanitizes data at ingestion, enforces entitlements at AI consumption, protects activity on embeddings in vector DBs, and inspects and controls prompts and responses.

3. Full Provenance of AI System:

A genAI system is often made up of a variety of building blocks and a myriad of complex relationships between ever changing data objects like files, user permissions, AI models, AI agents, vector databases, and user endpoints. It's important to have a full provenance view of the entire AI system, down to the level of each data object and file. Such visibility is also required by various AI regulations.

Gencore AI, powered by Data Command Graph uniquely provides the full provenance view of the entire AI System.

Therefore it is critical to have full visibility into provenance at a granular level. What data systems are feeding a particular LLM in Amazon Bedrock? Which files within this data system are being used? What users have access entitlements to these files? If I change a vector database in the system, what data systems are impacted? Gencore AI is powered by a unique knowledge graph that maintains granular contextual insights about data and AI systems. Not only does this support real time controls - it also provides comprehensive traceability of the entire AI system, including data and AI usage, down to the level of each file, user, AI model and usage end-points.

4. Compliance with AI Regulations for each AI System:

The incredible transformational power of generative AI has also propelled AI regulations in various regions and jurisdictions, such as EU AI Act and NIST AI RMF. There are dozens of other regional AI regulations being drafted globally. Organizations not only have to meet with base data protection regulations like GDPR for their AI Systems, but now also have to ensure compliance with new AI regulations.

Gencore AI uniquely provides compliance checks for each of the AI Systems being operationalized in it.