put·here
Put it here. Make sense of it later.
Pick AirDrop to send it to your Mac. No AirDrop? Email it to yourself or save it to Notes.
The phone app can’t import yet. Open this link on a Mac to put it in put·here.Skip the news cycle. These are the guides AI companies wrote themselves, plus a few books that will still hold up next year. Every card links to the original.
- Published: Portfolio, April 2024
- Why read it: Mollick teaches at Wharton and tests AI in real classrooms and companies. It is not a technical book; it is about working with AI as a colleague.
- 3 key points (three of his "four rules"):
1. "Always invite AI to the table." Try it on everything, because nobody yet knows where it helps or hurts in your field.
2. Be the human in the loop. AI does some tasks well but not whole jobs; judgment stays with you.
3. Assume this is the worst AI you will ever use. It improves fast, so don't write it off too early.
- Good for: anyone starting to use AI in daily work.
- Small print: key points summarized from the author's public descriptions and third-party summaries
- Sources: publisher page (official) https://www.penguinrandomhouse.com/books/741805/co-intelligence-by-ethan-mollick/ · four rules: Stanford GSB interview with the author, June 2024 (third party) https://www.gsb.stanford.edu/insights/co-intelligence-ai-masterclass-ethan-mollick
- Published: O'Reilly, December 2024
- Why read it: The whole craft of building products on top of foundation models in one book: evaluation, prompting, retrieval, agents, fine-tuning, cost and latency.
- 3 key points:
1. Decide how you will judge "good" before you tune anything. Open-ended output can be graded by another model.
2. Go from simple to complex: prompts first, then retrieval (RAG), then agents or fine-tuning.
3. Once live, latency and cost matter as much as quality when you pick a model.
- Good for: founders building AI products or working closely with engineers.
- Small print: key points summarized from the author's and publisher's public descriptions
- Source: O'Reilly book page (official) https://www.oreilly.com/library/view/ai-engineering/9781098166298/
- Published: O'Reilly, November 2024
- Why read it: A structured take on prompting, from the underlying ideas to concrete methods, instead of scattered tips.
- 3 key points:
1. A few good examples (few-shot) often beat a long list of rules.
2. Letting the model reason step by step (chain of thought) helps on hard problems.
3. Putting the right material into the prompt (RAG) is more reliable than hoping the model remembers.
- Good for: people who write prompts daily and want consistent results.
- Small print: key points summarized from the author's and publisher's public descriptions
- Source: O'Reilly book page (official) https://www.oreilly.com/library/view/-/9781098156145
- Published: December 2024 (byline: Erik S. and Barry Zhang)
- Key points:
1. Workflows follow steps you wrote; agents decide their own next step. If a workflow solves it, don't build an agent.
2. Common patterns: chaining, routing, parallel runs, an orchestrator with workers, and a writer–critic loop.
3. Three principles: keep it simple, make its planning visible, and design and test its tools carefully.
- Source (official) https://www.anthropic.com/engineering/building-effective-agents
1. Context is a limited budget. More is not better; include only what helps the task.
1. Agents fit work that needs judgment, has rules too messy to maintain, or lives in documents and conversations. If clear rules work, use plain automation.
1. Define success and a way to test it before you start editing prompts (Anthropic).
Where: Y Combinator AI Startup School, San Francisco, June 2025 (video on YC's official YouTube channel, uploaded June 2025)
For one week, try AI on everything you do and note where it helps and where it doesn't. (From: Co-Intelligence — Ethan Mollick)
Three lines on what a good result looks like, five real examples, and re-run those five every time you change the prompt. (From: AI Engineering — Chip Huyen;…
Give AI less but better: clear titles, short sections, all in one place. (From: Anthropic — Effective context engineering for AI agents; Andrej Karpathy —…
| # | Source | Type | Date | Checked |
|---|---|---|---|---|
| 1 | Penguin Random House — Co-Intelligence | Official (publisher) | 2024-04-02 | 2026-10-28 |
| 2 | Stanford GSB interview with Mollick | Third party | 2024-06-11 | 2026-10-28 |
| 3 | O'Reilly — AI Engineering | Official (publisher) | 2024-12 | 2026-10-28 |
| 4 | O'Reilly — Prompt Engineering for LLMs | Official (publisher) | 2024-11 | 2026-10-28 |
| 5 | Anthropic — Building effective agents | Official | 2024-12-19 | 2026-10-28 |
| 6 | Anthropic — Effective context engineering | Official | 2025-09-29 | 2026-10-28 |
| 7 | OpenAI — A practical guide to building agents | Official | undated | 2026-10-28 |
| 8 | Anthropic prompt engineering overview | Official | live docs | 2026-10-28 |
| 9 | OpenAI prompt engineering guide | Official | live docs | 2026-10-28 |
| 10 | YC video, Karpathy | Official (host) | uploaded 2025-06-18 | 2026-10-28, title and channel checked; length not verified |
| 11 | ikyle.me summary | Third party | 2025-06-21 | 2026-10-28 |
Note: the book key points are summarized from publisher pages, chapter lists and reputable summaries; we did not read every page of each book. peter can replace them with his own takeaways.