Updates

---

19 June 2026

Social AI Expedition coffee 4 · Den Bosch online — the first field note in English

• Social AI Expedition coffee 4 · Joost, Lars, Marijn van der Steen & Sameer Pashikanti · Den Bosch: a new report from a coffee conversation on 10 June, at the Dembrane office. Four voices, the conversation held partly in English. Read it at socialaiveldgids.nl/en/veldtocht-koffie-4.
• New "Field notes" section: the English field guide now has a "Field notes" section in the sidebar, collecting reports from coffee conversations around the Social AI Expedition. Coffee 2 · Utrecht and coffee 3 · The Hague are now available in English here too.

22 May 2026 (later that day)

Social AI Principles: nuance on principle 1 + a new sub-nuance on verbatim quotes

• Principle 1 rewritten to a two-layer structure. The old phrasing "AI adds nothing" was too black-and-white and clashed with other principles (scalability in principle 3, AI value levels, the Broaden role). New phrasing: "Wisdom lives in the system. AI helps make it visible." Layer 1 holds that the wisdom comes from the system itself (no external expert language as a starting point). Layer 2 acknowledges that AI legitimately adds things (structure, connections, hypotheses, scale, readability), as long as those additions are deliberately chosen and explicitly marked.
• New "Discipline-in-development" block as a foreword note to the whole bundle. It explicitly acknowledges that we are at the start of this AI journey of discovery. Rules are working hypotheses, not law. Marking AI inference ("this is an assumption the AI itself makes", "possibly underexposed") is valuable where it can be done, sometimes feasible, sometimes not. Room for what develops further.
• New sub-nuance to principle 4 (Ownership through language): "Verbatim is verbatim-in-meaning, not stenography". Filler words (uh, um), unintended stumble-repetitions and speaker-corrected false starts are not part of what someone "said", so cut them. Word choice, sentence structure, intentional repetition for emphasis, and characteristic speech or dialect you leave untouched. The test: would the speaker recognize themselves? No marking in the output (no [uh] or [...]). Practical occasion: a two-day session on 19-20 May where live quotes were verbatim-correct but unreadable when read aloud to the room.
• Language refresh: the word "wiki" replaced with "AI output" where it was meant generically, not specifically about a wiki application.

Read it at socialaiveldgids.nl/en/bedachtzaam/principles.

22 May 2026

Field notes section + first report online

• New "Field notes" section: in the sidebar, between Resources and Extra. This is where reports of coffee moments and other conversations around the Expedition land.
• Social AI Expedition coffee 2 · Joost and Lars · Utrecht: the first report is live. Six chapters from a recording on 13 May: the check-in, the break, the mirror, the paradox, the campfire and the echo. Read it at socialaiveldgids.nl/en/veldtocht-koffie-2.

21 May 2026

Thoughtful with Social AI

• New section in the sidebar, under Start. It collects the principles beneath the work: Social AI Principles, Prompt Best Practices, Ownership, and Bottom-up. Plus a Complete Bundle (AI input) for anyone who wants to hand the whole framework to a model at once.

3 March 2026

German translation available

• Full translation — All 19 pages of the field guide are now also available in German: Phase 1, Phase 2, Phase 3, and the main pages. The field guide is now available in three languages: Dutch, English, and German. You can switch languages via the language button at the bottom of the sidebar.

2 March 2026

English translation available

• Full translation — All 19 pages of the field guide are now available in English: Phase 1, Phase 2, Phase 3, and the main pages. You can switch languages via the language button at the bottom of the sidebar.

27 February 2026

Phase 3 begins — first deep dive online

• Phase 3: Scale — Main page expanded with the skeptic story, the music metaphor, four organizational patterns, and six lenses. The shortened version has been replaced with the full introduction.
• Patterns over time — First Phase 3 technique published. From snapshot to trajectory: the Gerda pattern, the two-step method (first separately, then together), and the full prompts for Transcript Analyst and Predictive Synthesizer.
• Two further deep dives (finding hooks, ownership) coming soon.

23 February 2026

Prompt rendering improved

• Tooltips for heading prompts — Prompts with # headings now show explanation tooltips from the "Why this structure works" section. Previously, tooltips only worked with bold labels.
• Compact headings — Heading levels in prompt blocks are now visually more compact and distinctive
• Markdown rendering — Improved display of bold, italic, and heading markers within prompt blocks
• Blockquotes — Quote lines (>) in prompts are now rendered as styled quote blocks

18 February 2026

Two new tensions and a practical tip

• Phase 2: Deepening — Added tension about how you introduce AI to a group: framing (experiment vs normal part of the session) and the shift from judging to recognizing
• Live reflection with AI — Practical tip: prepare prompts in advance, but know you'll need to adjust them when the session goes in a different direction
• Safe practices — Dembrane privacy level updated to high-medium (built for facilitation, privacy-focused, European servers)

13 February 2026

Launch — version 1.0

• Social AI Field Guide live — The visual platform of the field guide went live. This marks version 1: Phase 1 (transcription, language as ownership, source document style) and Phase 2 (live reflection, iteration, intuition in writing, and more) are available. The content was previously available on GitHub as Social AI Field Guide.