Delivering enterprise agentic solutions at scale
summary
Prototype becomes an enterprise Reed.AI MVP that handles 50 million CVs with a 1,000‑concurrent‑user SLA, built on AWS with RabbitMQ and EKS, OpenAI foundational models, Terraform + ArgoCD, and ISO27001/cyber essentials plus controls. Claims: two‑stage anonymization and ESCO ontology mapping protect PII while preserving matchability; matching runs a transparent algorithm (not an LLM); costs and risks (≈$0.05 per LLM call, denial‑of‑wallet) force event‑driven, GitOps, and agent‑based context control.
AI-written summary · about
chapters (18)
- 0:00Enterprise agentic AI solutions
- 1:32Consultancy phase
- 3:02Reed AI product
- 4:49Technical architecture
- 6:12Governance challenges
- 8:39Frontier technology
- 10:04AI vs engineering skills
- 11:30Prototyping architecture
- 13:18Conversational interface
- 16:46Data extraction & PII
- 19:38CV processing pipeline
- 22:05Context & flow control
- 24:13Scaling production
- 26:00Kubernetes adoption
- 28:06Security risks
- 30:19Key takeaways
- 31:42About Tangent
- 32:46Q&A
related talks
40mRolling out Enterprise AI: Tools, Insights, & Team Empowerment
58mCognizant cognitive architecture for generative AI: The path to MVP
33mMaster enterprise-scale generative AI
20mExplore what’s possible with new AWS generative AI services
46mBecoming an AI engineering company
34mOrganized chaos adopting GenAI at Enterprise
01h 42mBuilding Responsible AI Agents in Decentralized Open-Source Environments
11mOpen source GenAI vs. enterprise GenAI: Navigating the future of AI innovation