A practitioner's account of the year generative AI re-platformed the creative industries — written from inside the work, week by week, between October 2025 and May 2026.
📖 Read the latest PDF (21 May 2026) →
Each rebuild of the book ships as its own dated PDF, so earlier editions stay readable alongside the current one. The newest edition is always at the top.
| Edition | Newsletter span | |
|---|---|---|
| 21 May 2026 (current) | Issues 1–30 | Dream_Machine_2026-05-21.pdf |
| 14 May 2026 | Issues 1–29 | Dream_Machine_2026-05-14.pdf |
This is not a static manuscript. Dream Machine is a living book, rebuilt weekly from the Dream Machine newsletter and its evolving research archive. Each week a new newsletter issue ships, the corresponding chapters are updated, footnotes are added, predictions are checked against the moment, and the PDF is re-rendered.
If you've read it once, the version you read next month will be different — sometimes by paragraphs, occasionally by chapters. The book is a snapshot of a transition in motion. That is, in part, the point.
- Dream Machine newsletter — the weekly LinkedIn newsletter that the book is built out of. New edition every week.
- Dream Machine podcast — long-form conversations on the same material.
- DreamLab AI Collective — the studio in the North West of England where the work is done.
Written by Pete Woodbridge — creative technologist, founder of DreamLab AI Collective.
Eighteen chapters and eight deep-dive appendices, ~160,000 words, covering:
- The Tilly Norwood week and the launch of Sora 2
- The Human–AI Agency Continuum (a practitioner's framework)
- The Dead Internet / Living Web split
- The Slop Ceiling and the Authenticity Premium
- The UK's 88% copyright consultation and the Petrillo-template levy mechanism
- Four strategic positions the studios are choosing between
- World models replacing flat video as the default medium
- The Orchestrator role and the AI Literacy Premium
- The Age of the Why — the chess-grandmaster strategy applied to creative work
- A five-year speculative future-cast (Chapter 17)
- A complete inventory of every significant tool, platform and model from the period
See the full table of contents in the Foreword.
Book/
├── 00_Foreword.md ... 18_Epilogue.md # The chapters
├── A1_…md ... A8_…md # The deep-dive appendices
├── assets/
│ ├── cover.png, back_cover.png # Cover art
│ └── book.css # Book typography
├── build/
│ └── Dream_Machine_YYYY-MM-DD.pdf # Dated edition (one per rebuild)
├── build_book.py # The build pipeline
└── watch_book.py # Auto-rebuild on file change
Dream Machine MD/ # Newsletter issue archives
Dream Machine PDFs/ # Newsletter PDF archives
Deep Dive MD/ # Deep-dive research source material
Research/ # Underlying research notes
Requirements: Python 3.11+, Pandoc, and Microsoft Edge (Windows). One-time setup:
pip install pymupdf
Build the PDF:
python Book/build_book.py
The build is a two-pass pipeline — Pandoc converts the chapter markdown to HTML, headless Edge renders that HTML to PDF, drop-cap positions are scanned to discover chapter page numbers, and the table of contents is re-rendered with those numbers before a final merge with the standalone cover PDFs.
Each build writes to Book/build/Dream_Machine_<EDITION_SLUG>.pdf where EDITION_SLUG is set near the top of build_book.py. To cut a new dated edition, bump DATE and EDITION_SLUG together and re-run — prior editions are left untouched on disk so older readings stay reachable.
Auto-rebuild while editing:
python Book/watch_book.py
If you don't want to run the toolchain, every dated edition is committed in Book/build/ — open the most recent one (currently Dream_Machine_2026-05-21.pdf) or any prior edition from the table above.
The newsletter has only ever been as good as the community of creatives, technologists, union reps, academics, festival programmers, indie filmmakers, working musicians and audience members who have, week after week, sent in the things the editor would otherwise have missed. If you've got something we need to know about for the next edition, reach out via the newsletter or DreamLab.
Dual-licensed at your option under either of:
- MIT License (LICENSE-MIT or opensource.org/licenses/MIT)
- Apache License, Version 2.0 (LICENSE-APACHE or apache.org/licenses/LICENSE-2.0)
Unless you explicitly state otherwise, any contribution intentionally submitted for inclusion in this work by you, as defined in the Apache-2.0 license, shall be dual-licensed as above, without any additional terms or conditions.
© 2026 Pete Woodbridge / DreamLab AI Collective.