Agents

Feature: Agentic features open the door to data exfiltration or worse

Commentary on Agentic AI and Security by Stephen Downes. Online learning, e-learning, new media, connectivism, MOOCs, personal learning environments, new literacy, and more

Stop wasting time on 20-minute builds that do the wrong thing. Compyle keeps you in the driver's seat—planning together first, then asking when anything's unclear.

The race between human-centered work and infinite PowerPoints

A new generation of apps you can chat with and the tools for developers to build them.

The Rundown: In this guide you'll learn how to build a content brainstorming app using Google's Opal which can help you turn blank page syndrome into

Episode 76: What actually makes something a real "AI Agent"—and how close are we to AI handling complex work entirely on its own?*Want our guide to master AI...

Stop shipping agents like they’re apps. They’re self-governing code touching live systems. IBM × Anthropic is the signal: enterprise agents are here, and governance just became table stakes. Here’s the playbook I’m seeing (10 pieces you can ship, certify, and scale): ✅ what are ai agents? adaptive systems that reason, act, and learn with tools, not static chat apps. ✅ agentic enterprise embed agents into ops so decisions, workflows, and automations improve every run. ✅ ADLC (agent development lifecycle) devsecops for agents: design → sandbox → red-team → certify → deploy → monitor → retire. ✅ enterprise considerations tie use cases to ROI, controls, and regs; write the business case before the prompt. ✅ observability & ops beyond uptime: track behavior drift, tool errors, chain depth, reasoning quality, and rollback readiness. ✅ agent security defend against prompt injection, data leakage, privilege escalation; least-privilege tools with signed calls. ✅ governance: test, certify, catalog treat agents like services: pre-release evals, attestations, lineage, and an internal marketplace. ✅ MCP servers lifecycle model context protocol as a first-class surface: auditable, scoped, and versioned integrations. ✅ reference architecture hybrid stack that separates knowledge (RAG), capability (tools), policy (guards), and memory (state). ✅ voice of the customer & use cases ship real deployments (healthcare, telecom, finance) with before/after metrics, not vibes. If you’d word any of this differently, I’m all ears, drop your version and I’ll pin the clearest take. IBM/Anthropic folks, feel free to sharpen this for the operators in the trenches. Bottom line: this isn’t “labs” anymore. If you can’t test it, certify it, and roll it back in minutes, you shouldn’t run it in production. Follow Alex for operator-grade AI agents you can copy, and repost to put this in front of one teammate who owns your next deployment. Thanks Andreas Horn for sending this over. | 51 comments on LinkedIn

In this course taught by Andrew Ng, you'll build agentic AI systems that take action through iterative, multi-step workflows.

The race between human-centered work and infinite PowerPoints

: Analyst predicts over-supply will trigger a market correction in favor of deep-pocketed incumbents