Why this is in the vault
Every's weekly "Context Window" Sunday roundup recaps three Vibe Checks and several items already filed to the vault under their own dated notes — the one genuinely new item is Mike Taylor's "Good Writing With AI Starts Before the Prompt," a 12-step human-bookended writing process not previously captured.
Mapping against Ray Data Co
The concrete connection is to RDCO's own writing production discipline (feedback_writing_standard_contract, the WRITING-rdco.md contract, and feedback_ic_vs_production_mode's 12-stage production workflow for public-facing content): Taylor's process is a comparable structure worth checking RDCO's pipeline against — human judgment front-loads the work (title, sources, spitball via Monologue) before a model drafts, then human authorship returns for the ending and editorial debate, with the model doing only the middle draft. That's a specific, adoptable "human bookends, model fills the middle" pattern to test against where Sanity Check and X drafts currently let the model run start-to-finish.
Issue contents
- Vibe Check — Opus 5.5 pulls Codex converts back to Claude (Katie Parrott): already vaulted at [[2026-09-22-every-vibe-check-opus-5-5-codex-converts]].
- Vibe Check — GPT-6 Sol vs. Opus 5.5 (Dan Shipper): already vaulted at [[2026-09-22-every-gpt6-sol-vs-opus-5-5-vibe-check]]. This issue adds Shipper's one-line split verdict: Sol for a fast, cheap daily driver that leads with the point; Opus 5.5 for ambitious coding/visual work worth paying more for.
- Mini-Vibe Check — Grok 4.7 (Laura Entis): won back one power user (Tyler Nishida), not the rest of the team. Already vaulted inside the curation section of [[2026-09-24-every-why-evals-are-so-hot-right-now]].
- Steal This — "How to Create Your Own Personal AI Benchmark" (Mike Taylor): already vaulted at [[2026-09-21-every-personal-ai-benchmark]].
- Steal This — "Good Writing With AI Starts Before the Prompt" (Mike Taylor) — NEW, not previously filed: Taylor's 12-step process compresses a day of writing into under two hours by thinking before prompting — settling a title, gathering sources, and spitballing in Monologue — before a model drafts; after picking the best parts across draft versions, he writes the ending and debates the finer points with editors himself, keeping human authorship at both ends of the pipeline while the model fills the middle. Every also filmed Taylor, Katie Parrott, and Jack Cheng writing/editing on camera live as the closer to its "How We Write Now" series.
- Steal This — "How to Get the Most Out of Jev" (Laura Entis): already vaulted at [[2026-09-23-every-jev-usage-guide]].
- Steal This — "Turn an Article Into a Video" (Becky Isjwara's storyboard-to-Hyperframes workflow): already vaulted inside [[2026-09-24-every-why-evals-are-so-hot-right-now]].
- Knowledge Base — "Copilot Gets a Seat in the Org Chart" (Ryan Sloan): already vaulted at [[2026-09-25-every-copilot-org-chart-autopilot]].
- Thesis Statements — five new entries (Drew Breunig, Sean Campbell, Geoffrey Litt, Daniela Rus, Arielle Shipper): already vaulted inside [[2026-09-24-every-why-evals-are-so-hot-right-now]].
- Log On — Oct 2 prompting workshop (Micah Rich) and Thesis: 2027 ticket countdown: subscription/event CTAs with no extractable technique — triage-skip per the standing Every live-session/Camp promo rule.
⚠️ Sponsorship
No third-party paid sponsor in this issue. Every runs its standard house footer — a "Want to sponsor Every?" CTA plus the bundled self-promo block for Every All Access and its own products (Sparkle, Cora, Spiral, Monologue). Flagged per standing practice (sponsored always needs a paired sponsor_entity); it does not bias any of the reported content above.
Related
- [[2026-09-22-every-vibe-check-opus-5-5-codex-converts]]
- [[2026-09-21-every-personal-ai-benchmark]]
- [[2026-09-24-every-why-evals-are-so-hot-right-now]]
- [[feedback_writing_standard_contract]]
- [[feedback_ic_vs_production_mode]]