Floki: Defining Agentic Workflows for Security
summary
Floki is an open source Python framework built on Dapr sidecar that defines agentic workflows as code, turning LLM-driven steps into microservices with built-in telemetry, Redis state, and publish/subscribe. Demoed pieces include an AI-generated podcast pipeline that downloads ArXiv papers, summarizes pages with an LLM, and uses 11Labs TTS (two episodes published), and an incident‑triage multi‑agent workflow that indexes schemas, generates graph ontologies and Cypher/Neo4j queries. Runs locally in Docker, mostly works but demos show occasional task failures and retries.
AI-written summary · about
chapters (9)
- 0:00Introduction
- 3:42Agentic workflows
- 10:31Dapper workflows
- 24:03AI security podcast
- 32:23Cyber security challenges
- 42:18Multi-agent system design
- 57:18Flocky features & workflows
- 1:01:32Incident triage demo
- 1:08:38Q&A
related talks
35mFloki: Building Agentic Workflows with Dapr
11mFlowco: Mixed-Initiative Authoring of Reliable End-to-End Data Analyses via Dataflow Graphs and LLMs
59mOptimizing Complex Workflows with Event - Driven - Multi Agentic Approach
34mOrchestrating Generative AI Workflows to Deliver Business Value
28mAn introduction to the Workflow Development Kit
40mArchitecting Secure Agentic Workflows on Kubernetes: A Financial Sector...
30mBuilding Robust, Secure LLM and Agentic AI Workflows
31mArchitecting Secure Agentic Workflows on Kubernetes: A Financial Sector…