As AI changes the world in ways we never imagined, it’s easy to get overwhelmed with the possibilities and lost in the details. But with the trusted infrastructure to power and secure AI, Cisco can help you maximize AI’s value for your organization.
Ai4 2024, the biggest AI industry event in North America, will bring together thousands of technology innovators and executives in Las Vegas. We’re thrilled to have four Cisco AI experts speaking on how we’re connecting and protecting organizations in the AI era:
- Delivering and Governing Trustworthy GenAI Capabilities at Scale
- Tushar Agrawal – Sr. Director, Product Management, GenAI, Outshift by Cisco
- August 13 | 11:35–11:55 a.m. | Room: 150-151
- Using AI to Enhance Observability and Business Resilience
- Jeff Wiedemann – Director, Technical Marketing Engineer, AI, Splunk, a Cisco Company
- August 13 | 12:25–12:45 p.m. | Room: 169-170
- Developing AI Intuition: From Model Basics to Chat GPT
- Chris Rowen – VP, Engineering Collaboration AI, Cisco
- August 14 | 10:35–10:55 a.m. | Room: 119
- AI-First Approach to Security
- DJ Sampath – VP, AI, Security Business Group, Cisco
- August 14 | 11–11:20 a.m. | Room: 111
In prep for the big event, we sat down with these experts for a behind-the-scenes look at what they’ll cover in their sessions and how Cisco is driving AI forward.
AI for faster detection and response: Jeff Wiedemann’s take
Jeff Wiedemann is a Technical Marketing Engineer at Cisco, focused on AI for Splunk. Jeff, a self-proclaimed nerd, works to ensure that Cisco’s products meet customer needs and are effectively communicated to the market. We talked to Jeff to learn more about the role AI plays in enhancing observability.
How does AI help with monitoring systems and fixing issues faster?
Jeff: As businesses grow, their environments become more complex—more services to monitor, more data to analyze, and more potential failure scenarios. This leads to tool sprawl and blind spots, which can make monitoring and responding effectively a real challenge.
Splunk’s AI-driven solutions suggest solutions you might not have considered, and offer next steps based on real-time data analysis. This means you can quickly detect and resolve problems.
Our AI approach is tailored specifically to your domain and takes full advantage of Splunk’s security and observability capabilities. We always keep a human in the loop to enhance decision-making, ensuring that AI supports rather than replaces human judgment. Plus, our platform is open and flexible, so you can customize and extend AI models to fit your unique needs. By using advanced AI, Splunk helps you cut through the noise to find the signals that matter so you are better equipped to withstand and quickly recover from digital disruptions.
What’s the best way to start integrating AI-driven solutions into your ecosystem?
Jeff: Start small and focus on specific pain points rather than trying to overhaul your entire system at once. Look for AI solutions that are ready to use, tailored to your industry, and easy to extend. These can solve common issues and provide immediate value without requiring a dedicated data science team.
For example, you could improve your incident response by moving toward a more proactive approach. Quick wins might include things like using machine learning (ML) for anomaly detection and analyzing historical trends to make your alerts more accurate. Using solutions with advanced AI and ML built in throughout the incident response workflow, such as Splunk Observability, can help you reduce your MTTD and MTTR in every step.
Deploying GenAI: Tips from Tushar Agrawal
Tushar Agrawal is the Senior Director of Product Management for Generative AI at Outshift, which is Cisco’s incubation engine that turns bold ideas into tomorrow’s tech. While Cisco connects and protects the world’s tech, Outshift explores emerging technologies, creating powerful solutions for IT visionaries in GenAI, quantum computing, cloud native, and beyond.
What are some best practices or things to watch out for when deploying GenAI technologies?
Tushar: According to the 2024 Cisco AI Readiness Index, 61% of companies believe they have one year or less to implement their AI strategies before facing negative business consequences from falling behind. To stay agile, it’s important to start small and scale quickly. Avoid getting bogged down in endless planning; instead, initiate a pilot project to gauge the waters. This allows you to determine what works and what requires adjustments. It’s also important to evaluate multiple models because different large language models (LLMs) can be better suited to different tasks. Experiment with multiple models to find the best fit for your needs.
How can IT and business teams speed up the delivery of GenAI capabilities to maximize value?
Tushar: Begin by initiating pilot projects to test and refine your approach. Share your successes across the organization to inspire others and foster a culture of GenAI innovation. Regular showcases that involve various departments can spark new ideas and collaborations, leading to innovative business practices. This way, you can scale quickly and effectively while keeping everyone motivated and engaged.
Driving clearer speech, audio, and visual experiences: Chris Rowen’s vision
Chris Rowen is a Silicon Valley entrepreneur and technologist known for his groundbreaking work developing RISC microprocessors, domain-specific architectures and deep learning-based software. As Vice President of Engineering for Webex Collaboration AI, Chris leads an engineering and product team focused on building smarter, clearer and more seamless speech, audio, visual and relationship intelligence experiences.
How do you understand and apply AI techniques to natural language, audio, and video?
Chris: You build an intuitive understanding by developing and deploying AI solutions. Not only does Cisco Collaboration have deep experience in building its own world-class audio, video, speech, and language models from scratch, but our AI team has worked with every product line in Collaboration to help engineers grasp the underlying science and to navigate the critical application choices.
By paying attention to these three core principles, engineering teams can make widely productive customer use-cases, not just showy demos:
- Start with quantitative targets of success in customer experience.
- Ensure the data used to train and test the AI model fully represents the users you serve.
- Iterate with the models and the application of models until the behavior doesn’t just make a flashy first impression, but actually fulfills the customer’s need.
What are the risks of using LLMs like ChatGPT, and how can they be managed in team communication and collaboration?
Chris: LLMs are remarkably versatile and clever, but they are trained more to deliver a plausible answer than necessarily to provide correct results for YOUR query.
Their statistical models naturally gravitate towards conventional, boring prose responses. The quality of answers is only as good as the inputs data they’re given, so if under-supplied with facts, they invent new ones. They can propagate the bias of the data they’re built with, putting some groups of users at a systematic disadvantage. If improperly managed and protected, they can leak private data to the model host.
The superficial brilliance of LLM responses can lull users into excessive confidence in model infallibility, unless the LLM is carefully prompted, fine-tuned, and monitored.
AI-driven security innovations: DJ Sampath’s perspective
DJ Sampath is a software engineer turned founder and CEO. He co-founded and was the CEO of Armorblox, which was later acquired by Cisco. He also helped found StackRox (now RedHat), where he served as Chief Architect and VP of Engineering. He now serves as Cisco’s VP of AI Products.
How is Cisco using machine telemetry at scale?
DJ: Cisco is using machine telemetry at an unprecedented scale to gain AI-driven insights. We have a “security meets the network” advantage, which gives us access to billions of endpoints and devices and provides an unparalleled foundation for capturing real-time data from applications, networks, security events, and the broader internet.
Using data from Cisco and Splunk, we process and analyze trillions of data points daily with AI and machine learning algorithms. This helps us proactively identify and mitigate security threats, optimize network performance, enhance application experiences, and better understand user behavior and trends.
How do AI-driven insights contribute to simpler security and better user experiences?
DJ: Cisco’s AI-first approach to security is transforming how we protect organizations and improve user experiences. We use AI to proactively detect and respond to threats early, simplifying security operations and reducing the risk of breaches. By analyzing trillions of data points, our AI algorithms spot anomalies and potential threats, allowing us to identify and counteract zero-day attacks before they can do any damage.
Our AI-powered tools speed up incident response by automatically correlating security events and triggering pre-defined actions. This makes security measures seamless and unobtrusive, which improves the user experience without compromising safety.
What innovations are being enabled through the integration of AI in security?
DJ: Our commitment to AI innovation is delivering cutting-edge solutions in behavioral analytics, anomaly detection, threat intelligence, automated response, and natural language processing in response to constantly evolving threats. The bad actors never stop, so we’re always vigilant and proactive in developing technologies and solutions that adapt and improve continuously. Ultimately, this enhances security posture and improves user experiences in an integrated manner.
Meet the experts and learn more
Attend our sessions at Ai4 to hear from our experts live. Visit our two booths (Outshift in #220 and Splunk in #609) to see Cisco AI solutions in action. We can’t wait to see you there!