Topic
Coding Agents - Page 4
Practical shifts in agentic coding, review, and delivery. Curated tldw.news briefings about coding agents, with practical engineering takeaways from long-form AI and developer-tool videos.
199
breakdowns
Page 4 of 20
Theo - t3․ggSmarter models just changed who gets to build software.
One developer rewrote the TypeScript compiler in Rust in under 20 hours. The real shift: smarter models let non-coders ship software they don't understand.
AI EngineerMonzo's Slack Agent Writes 10% of PRs. The Platform Did That.
Monzo's Slack agent authors 10% of merge PRs. The real story is the platform, MCP gateway, and sandboxing that made it safe.
AI EngineerAgent Harnesses Stop Adding Coding IQ. Now They Manage Cost and Risk.
Agent harnesses no longer make coding models smarter; they now shape cost, permissions, and which agentic work is possible. The shift is underway.
AI EngineerModel Swapping Only Works When Evals, Not Benchmarks, Decide
Four frontier models fail differently on the same tasks. Microsoft's model-swap workshop shows why custom evals, not benchmarks, should decide which LLM ships.
IBM TechnologyAgent loops run by themselves; comprehension debt is the new bottleneck
Loop engineering automates prompting, but agents still need human goals, documented intent, and review. Comprehension debt is the part most teams will miss.
InfoQLetting AI approve low-risk PRs cut Duolingo's median merge time
Duolingo's LLM bot auto-approves low-risk PRs, cutting median merge time from 18 to 12 hours. The hard part wasn't the model.
Hugging FaceDurable agents are coming. P Durable is not ready for your stack
P Durable is not a coding agent; it is a crash-resistant harness for building always-on agents. The demo works, but production durability remains unproven.
Theo - t3․ggMost features are prompts once you expose the data
A T3 Code maintainer argues most planned features should be prompts. His examples are concrete; the vendor bias is real.
AI EngineerAnthropic Opens Multi-Agent Orchestration to Every API User
Anthropic launches Flows for Claude Managed Agents: up to 1,000 self-orchestrated sub-agents via API. Cuts through agentic laziness but raises governance…
Cole MedinThe AI Dark Factory Dream Is Dead—Human Oversight Wins
Full autonomy AI coding fails. A human-in-the-loop software factory with scoped epics and holdout tests is the practical alternative.
Get TL;DW
Too Long; Didn't Watch.
A concise breakdowns of the AI and devtools videos that actually matter for engineering leaders.
Free. Weekly. No hype.