Elastic Austin Meetup - May 2019
Join the Elastic Austin User group: https://www.meetup.com/Austin-Elastic-Fantastics/
Elastic Bots: Analyzing Conversational AI for Artificial Capability Equivalence, Cognitive assistants, virtual agents, and chatbots have taken the world by storm and are now making their way into the large enterprise space. AI and machine learning initiatives are hot on every CxO ticket for 2019, but most organizations are unsure how they should measure the success of their investment and its effectiveness on the enterprise. We've built a data tagging pipeline using Elasticsearch + Kibana + Python that allows you to see real-time results from your bot deployment and provide meaningful (and beautiful!) reports to your stakeholders. I will focus on the ACE Scoring Method and Elastic Bot architecture, and then show live demos of conversations, data tagging, visualizations, and utterance extraction for continued learning.
?Patrick Marlow is the Director of Engineering and Chief Cognitive Architect at IPsoft focused on delivering Intelligent Automation and Cognitive Agents to the enterprise space. He has over 15 years of experience designing and crafting solutions using AI, automation, machine learning, and data analytics. When he's not trying to grok the latest NLP white paper or tune hyperparameters in TensorFlow, you can usually find him sending a gnarly V5 at Austin Bouldering Project, running around Zilker with his Australian Shepherd, or spending time outdoors with his wife and son.
Problem: Multitenancy – balancing security and stability, with empowerment and self-service
The current production cluster is maintained by a single team and all changes must go through a single person. This cluster was initially deployed to manage Information Security and compliance-related data and as such, the security setup is such that teams really can’t self-serve. We’re in the process of deploying a new, “Enterprise” cluster which has been architected to scale out, provide secured access to all teams across the organization, with varying use-cases and retention requirements. We would like to focus on how we’ve designed our deployment to use Tags, Security, Alerting, and Spaces.