Enterprise AI and workflow software
ServiceNow
ServiceNow builds enterprise software for managing and automating work across IT, customer service, human resources and security. Its ServiceNow AI Platform combines business applications, data and AI agents: teams can handle incidents, fulfill requests, onboard employees and remediate vulnerabilities. With AI Agent Studio, users can customize agents through natural language, define their roles and guardrails, and equip them with scripts, workflow actions and other tools.
Founded in 2004 by Frederic B. Luddy, ServiceNow grew from his goal of simplifying everyday work through software. Bill McDermott, who joined in 2019, is chairman and CEO. Its agent architecture connects planning to execution: AI Agent Orchestrator coordinates teams of agents, while support for the Model Context Protocol and Agent2Agent protocol lets them access external tools and communicate with agents outside ServiceNow.
Workflow Data Fabric supplies business context by connecting data across systems, organizing it through a unified catalog and applying policy-based governance. Operations managers, analysts and service representatives can use that connected information without switching systems—for example, bringing billing, support and inventory histories into customer issue resolution. AI Control Tower provides a central place to manage, monitor and govern AI, including third-party agents, alongside the workflows in which it operates.
Explore the recordings
This archive guide is separate from a company history: it covers the single ServiceNow recording in the supplied current catalog. The catalog establishes the recording’s title, speaker label, organization tag, link, and summary; technical and product-related statements below are presented as claims made in the recording, not as verified facts about ServiceNow’s current products or capabilities.
Start here: diagnosing voice-agent failures
Watch “My name is... my name is...”: A Linguistic Map for Voice Agents — Midam Kim, ServiceNow. In the recorded presentation, Midam Kim uses a failed voice interaction to organize problems across listening and speaking, then across sounds, words, interaction, and the user’s mental model. This is the archive’s broadest entry point for understanding how recognition, pronunciation, vocabulary, turn-taking, context, intent, and task completion interact.
Path for conversation and voice-system designers
Use the same Midam Kim presentation as a design-oriented route through turn detection, interruption, latency, shared vocabulary, context retention, emotion handling, and adaptation to different users. These are recommendations and relationships described by the speaker in the recording; the catalog does not establish that they represent current ServiceNow product behavior or implementation.
Path for evaluation and failure analysis
For an evaluation-focused viewing, follow the recording’s progression from individual speech errors to interaction failures, escalation, abandonment, failed tasks, and silent failure. The speaker identifies EVA Bench as an end-to-end diagnostic benchmark in the talk. That is a recorded claim; no current availability, ownership, implementation, or product status for EVA Bench is established by the supplied catalog.
Path for linguistics and user mental models
View the recording through its linguistic framing: speech disappears as it is produced, while the user’s mental model accumulates across the interaction. The speaker argues that this mental model should be a design target and closes by asking whether voice systems can adapt as users learn system behavior and language changes. Treat these as the presentation’s framework and conclusions rather than independently verified present-day facts.
1 talk
Newest first
1 speaker at AIE
Affiliations reflect their AIE appearances, not necessarily current employment.