A wave of companies is pushing AI agents beyond chatbot interfaces into practical, autonomous tools that can execute tasks across industries. Notion’s AI lead Sarah Sachs is building agents that manage inboxes, run meetings, and coordinate projects across Slack, GitHub, and Google Drive, with early testers creating over 21,000 Custom Agents before a public launch.
Dropzone AI’s CEO Edward Wu is targeting cybersecurity with an agentic security operations center that uses LLMs to triage attack alerts and investigate threats, mimicking human analysts without outsourced labor. Over 300 companies already use its system.
Fireworks AI’s Lin Qiao is tackling the high cost of running AI agents, processing over 15 trillion tokens daily on its inference platform. She advocates for smaller, customized models that boost efficiency 5 to 10 times, already deployed by more than 10,000 firms including Cursor, Uber, and DoorDash.
Tomorrow.io’s Shimon Elkabetz is coupling weather satellites with agentic AI to help clients evaluate storm risks, from drug delivery logistics to deploying snowplows. The platform maps storm shapes against company assets and analyzes cascading disruptions across travel networks.
At Anthropic, philosopher Amanda Askell is embedding ethical instincts into Claude for an agentic era where models make autonomous decisions—like managing investment portfolios—without human oversight. She focuses on making Claude responsive to individual user values rather than imposing its own ethics.
Counter-argument: Critics warn that delegating consequential decisions to AI agents risks amplifying opaque biases, security vulnerabilities, and accountability gaps, especially as these systems operate with increasing autonomy across sensitive domains like cybersecurity and finance.
Ai context: This brief synthesizes five Fast Company articles published within an hour of each other, each profiling a company or leader advancing AI agents. The sources are verified and highly relevant, but the brief cannot confirm the technical claims or usage metrics independently; all figures come directly from the articles.