# Week in review

Every Field Note grouped by the Monday–Sunday week it reports on, then by topic: 21 weeks, 4270 notes. JSON: https://newruntime.com/weeks.json.

- [Nov 16–22, 2026](https://newruntime.com/weeks/2026-w47.md): 258 notes; most covered: Models 31, Claude 29, Inference 17
- [Nov 9–15, 2026](https://newruntime.com/weeks/2026-w46.md): 570 notes; most covered: Models 73, Claude 71, Google 49. Agent infrastructure is getting cheaper and more controllable, while its security perimeter keeps expanding
- [Nov 2–8, 2026](https://newruntime.com/weeks/2026-w45.md): 661 notes; most covered: Claude 78, Models 47, Agent economics 43. The coding-agent loop is now a product to buy and cap, while agents reach money and still miss most real work
- [Oct 26 – Nov 1, 2026](https://newruntime.com/weeks/2026-w44.md): 851 notes; most covered: Claude 114, Models 56, Agent security 48. Agent adoption is moving into core software work faster than the controls that make it safe to scale
- [Oct 19–25, 2026](https://newruntime.com/weeks/2026-w43.md): 690 notes; most covered: Models 72, Claude 68, Anthropic 35. Agents are moving from demos into durable work, but control and verification are becoming the price of autonomy
- [Oct 12–18, 2026](https://newruntime.com/weeks/2026-w42.md): 398 notes; most covered: Models 43, Generative media 23, Claude 21. Open models and agent infrastructure are making deployment easier, while oversight and security become the constraint
- [Oct 5–11, 2026](https://newruntime.com/weeks/2026-w41.md): 147 notes; most covered: Models 17, Claude 9, Generative media 8. Agent systems are shifting from open-ended generation toward explicit decisions, controls and usable work surfaces
- [Sep 28 – Oct 4, 2026](https://newruntime.com/weeks/2026-w40.md): 96 notes; most covered: Agents 11, Claude 6, Coding agents 6. Agent products are shifting from isolated tasks to persistent, governed work systems
- [Sep 21–27, 2026](https://newruntime.com/weeks/2026-w39.md): 108 notes; most covered: Claude 11, Agents 7, Models 7. AI agents are moving from capable interfaces to governed, cost-aware operating systems
- [Sep 14–20, 2026](https://newruntime.com/weeks/2026-w38.md): 24 notes; most covered: Security 7, Coding agents 5, Agent evals 2. Agent systems are becoming more operational—and harder to trust by default
- [Sep 7–13, 2026](https://newruntime.com/weeks/2026-w37.md): 42 notes; most covered: Agents 11, AI 9, Models 9. Enterprise AI is shifting from tool adoption to governed execution and measurable outcomes
- [Aug 31 – Sep 6, 2026](https://newruntime.com/weeks/2026-w36.md): 45 notes; most covered: Agents 20, Agent security 11, Coding agents 4. The bottleneck is shifting from making agents capable to making their work verifiable, controlled and portable
- [Aug 24–30, 2026](https://newruntime.com/weeks/2026-w35.md): 25 notes; most covered: Agents 11, Agent security 5, Agent evals 2. Agent work is shifting from prompted sessions to persistent, permissioned systems
- [Aug 17–23, 2026](https://newruntime.com/weeks/2026-w34.md): 48 notes; most covered: Agents 15, Models 5, Coding agents 3. AI work is becoming an operated, packaged system rather than a model call
- [Aug 10–16, 2026](https://newruntime.com/weeks/2026-w33.md): 74 notes; most covered: AI 15, Coding agents 7, Agents 6. Agent value is shifting from model choice to the runtime that routes, governs and verifies work
- [Aug 3–9, 2026](https://newruntime.com/weeks/2026-w32.md): 85 notes; most covered: Coding agents 15, Agents 10, Open models 5. AI agents are moving from clever sessions to governed, measurable operating systems
- [Jul 27 – Aug 2, 2026](https://newruntime.com/weeks/2026-w31.md): 90 notes; most covered: Coding agents 14, Agents 12, Agent evals 7. Agent work is moving from generating output to governing execution and verification
- [Jul 20–26, 2026](https://newruntime.com/weeks/2026-w30.md): 55 notes; most covered: Coding agents 6, Agent security 3, AI adoption 3. AI agents are shifting from clever sessions to governed, testable operating systems
- [Jul 13–19, 2026](https://newruntime.com/weeks/2026-w29.md): 1 notes; most covered: Memory 1. Useful short-term context can be written into model parameters, then used to generate further training.
- [Jul 6–12, 2026](https://newruntime.com/weeks/2026-w28.md): 1 notes; most covered: Inference 1. Speculative decoding becomes a runtime scheduling choice because a fixed depth can slow high-batch inference
- [Apr 6–12, 2026](https://newruntime.com/weeks/2026-w15.md): 1 notes; most covered: Agent identity 1. Agent identity is being recast as a short-lived delegation chain, not a reused login
