Andrew Ng
9 articles tagged with this topic
AI Projects Keep Failing — Stop Blaming LLMs, Engineering Is the Real Problem
Tech team post-mortem: not the model's fault — bad prompts, unsanitized input, unvalidated output. A systemic enterprise AI failure pattern.
Tool calling, not LLMs, is where 90% of AI Agent projects die
Tool calling is AI Agent's action layer: the model declares the function, code runs it. This plumbing decides whether enterprise AI is real—or just a
90% of AI Agents Are Wrappers — Three Tests to Filter Vendor Hype
Real agents have clear technical definitions. Three hard metrics from frontline practice to help buyers cut through vendor pitch in AI Agent selection
Andrew Ng Pivots: Coders May Not Be Future Winners
Andrew Ng just repositioned DeepLearning.AI around 'AI engineering' after 10,000 job postings. What does it mean for non-coder solopreneurs?
AI Automation's Most Dangerous Failure Isn't Errors—It's 'Fake Success'
When AI workflows report 'success,' the task may have silently failed. A developer's four safeguards offer a blueprint for any enterprise deploying AI
Andrew Ng Parses 10,000 AI Job Postings: Judgment, Not Coding, Is What's Scarce
Andrew Ng's team clustered 10,000+ AI job postings into 4 must-have AI engineering skills. The deeper finding: the role shifts from "coder" to "decisi
AI Edits Code, Tests Pass, Prices Wrong — An Invisible Bug Hits Enterprise IT
Andrew Ng's 90% statistic lands as a real case: Codex edited train ticket round-trip pricing code, tests passed green, yet prices quietly went wrong.
Mission Driver Writes AI Thinking to Files — A Chinese Dev's Autonomy Bid
AI assistants now run one task, then quit. Mission Driver externalizes AI reasoning to text files for week-long AI runs with mid-mission pivots.
Headroom's Reversible Compression Cuts Agent Token Costs Without Sacrificing Accuracy
Headroom's CCR reversible compression mechanism was dissected by the developer community this week: compress tool outputs, store them locally, let mod