We recently launched our flagship Generative AI (GenAI) platform known as Persistent GenAI Hub. This groundbreaking platform enables companies to create faster, more efficient, and secure GenAI experiences at scale, anchored by the principles of Responsible AI. Enterprises can adopt GenAI across different Large Language Models (LLMs) and clouds, with no provider lock-in, integrated with existing assets and enabled by pre-built accelerators. With the Persistent GenAI Hub, enterprises have a complete platform and roadmap for accelerated GenAI adoption and faster go-to-market for new AI-powered services.
Crew.AI is a cutting-edge framework for building LLM-powered agents. It enables the orchestration of domain-specific agents and role-playing so these agents can capture domain knowledge. Powered by best-in-class tooling, the agents collaborate to enable a specific use case following the principles of planning and reflection.
Here, we will show a pattern for integrating Crew.AI with the Generative AI Hub. This is a natural fit to combine the orchestration power of Crew.AI with the cross-LLM capabilities of the GenAI Hub, with a unified view on cost and observability, all controlled via Responsible AI guardrails. We will specifically show an AI Cybersecurity use case where we help a customer, a major wealth management firm, compile cyber threat intelligence feeds and align them with mitigations defined in their internal policy document.
The challenge faced by the customer was that there were so many sources of threat intelligence in formats like STIX, and they were available directly on websites like TheHackerNews. We created 3 agents using Crew.AI, a Senior Security Researcher that looks at the latest news feeds and scrapes recent news on a specific security topic provided by the user. Then, a Policy Analyzer agent accesses internal documents to understand what policies apply to which security alerts and mines the recommended mitigations. Finally, a Technical Writer agent combines all this knowledge into an intuitive article for consumption of an expert. All the LLM access and key management is provided by the GenAI Hub. Hence, we could try this multi-agent application on different models like Azure OpenAI GPT-4o, Claude3 Sonnet via AWS Bedrock and Google Gemini. For each configuration, we could drill down to the individual prompts being passed on to the LLM and this full observability helped us update our agent prompts. We could also clearly observe the cost associated with the entire agentic workflow we have defined.
The solution helps our customer to autonomously compile security information from multiple internal and external sources and provide a consolidated, actionable report. As we see more such use cases combining GenAI Hub with advanced frameworks like Crew.AI, we look forward to some powerful applications being developed for different domains, such as Banking, Insurance, Life Sciences, Healthcare, Telecom and Retail.
As Generative AI continues to trend, companies need expert guidance to maximize their GenAI investments. That’s where we come in. To learn more about Persistent.AI offerings, please reach out to us.
