ChatGPT Setup Guide v3.8, 6 September 2026
Updated the current-model section for the phased GPT-6 Astra rollout, separated regular Chat access from Work and Codex access, preserved GPT-5.6 fallback guidance, and added a model-lineup regression gate.
The AI Setup Guides ยท Changelog
Every edition, dated, with what changed.
The guides get re-checked against the live products and republished whenever the tools move. Every edition lands in the same library buyers already use, at the same address. This page is the public record of what shipped and when, so you never have to take the word "current" on faith.
Published editions
Edition 2026.4 is current. Claude is at v3.7, ChatGPT at v3.8.
An edition covers both guides. Each guide also carries its own version number in its footer, listed below so you can tell at a glance whether the copy you are reading is the current one. The two guides do not always share a version number: a change that only affects one of them only moves that one.
Edition 2026.4
16 July 2026, last updated 6 September 2026
Current
Beginner to AI Architect v1.3, 5 September 2026
Added a safer bounded AI workspace starter, separate personal, business, and client roots, clearer hands-on project language, honest lab timing, route-specific Fast Track progress, stronger Git and tool-containment exercises, and a more direct first-use sales path.
Claude Setup Guide v3.7, 5 September 2026
Updated the model lineup to Fable 5.1 and clarified current Fable plan access against Anthropic's official model and support pages.
ChatGPT Setup Guide v3.7, 26 August 2026
Rewrote the hallucination hunt to model checking the live Scheduled Tasks page and account screen instead of asserting a disputed plan cap.
Beginner to AI Architect v1.2, 26 August 2026
Added the Owner-Operator Fast Track from existing lessons, improved the first lesson's tool glosses, expanded MFA on first use, and replaced an out-of-level recall question.
Dual AI Prompt Vault v1.1, 5 September 2026
Added a three-minute quick start, explicit placeholder replacement and privacy guidance, a completed example, section jump links, and return-to-filter controls for long mobile sessions.
Claude Setup Guide v3.5, 12 August 2026
Corrected Max capacity language, Memory paths, Artifact sharing limits, Claude Code permission modes and surfaces, connector approval wording, and Cowork scheduled work. Added safer disposable-copy instructions and installable Skill guidance.
ChatGPT Setup Guide v3.6, 12 August 2026
Corrected Work and Codex plan and surface availability, clarified Skills, connectors, and Plugins, strengthened first-run permissions, and replaced sensitive financial exercises with downloadable synthetic fixtures.
Beginner to AI Architect v1.1, 12 August 2026
Added the AI Command Center, prerequisite-based setup bands, three early guided missions, safe coding and finance fixtures, exact instruction-file downloads, private draft controls, spaced review, resource search, lesson session estimates, and a transformation receipt.
Claude Setup Guide v3.4 and ChatGPT Setup Guide v3.5, 11 August 2026
Published the purchase-access, recovery, and buyer-isolated progress hardening release before the 12 August vendor fact pass.
Purchase access and refund protection, 11 August 2026
Clarified the 30-day satisfaction guarantee and personal licence. A successful full refund now ends guide access, active browser sessions, recovery access, and future editions. Partial refunds keep access active, and a later purchase restores the same saved progress.
Bundle library, 9 August 2026
Added the text-based orientation and a plain explanation of why Claude uses worked practice blocks while ChatGPT uses retrieval checks, with both formats feeding the same seven-day review queue. A drafted presenter video, poster, and captions were not published in the library.
Claude Setup Guide v3.3, 9 August 2026
Corrected provider sign-in, retention, Incognito, plan pricing, sensitive-data permission, and Cowork availability against Anthropic's live support pages. Added everyday-first path routing, a seven-day practice review queue, screen-drift support templates, buyer-isolated browser progress, accessible heading order, and core completion that does not require the optional operator section.
ChatGPT Setup Guide v3.4, 9 August 2026
Corrected the model lineup, provider sign-in, Temporary Chat, signed-document privacy answer, Work and Codex availability, Scheduled Tasks uncertainty, and the final answer key against OpenAI's live pages. Added the same everyday-first path routing, seven-day review queue, screen-drift support templates, buyer-isolated browser progress, accessible heading order, and optional operator completion rule.
Claude Setup Guide v3.2, 8 August 2026
Reordered so your first real win comes before the privacy setup instead of after it, and split privacy into two shorter sections. Rebuilt "It can be confidently wrong" as an exercise you do rather than a warning you read. Three new sections: knowing when a request is past what Claude can do, Claude Code now that it is a tab in the desktop app rather than a terminal, and what it costs including the shared usage pool. Rebuilt "Talk to it well" around four slots and a verification habit, and split it in two: building the request is now Section 12, grading the answer is Section 13. Added a personal and a business example to every section that had none, ten rendered practice boxes that hold their reasoning back until you have answered, side-by-side before and after pairs, and a closing line on each section naming what is next and how long it takes. Replaced the self-rating scorecard at the end with three checks on real output. Twenty-five numbered sections, 133 minutes.
ChatGPT Setup Guide v3.3, 8 August 2026
No facts changed in this one. It rebuilt the teaching. "Talk to it well" is now built on four slots (outcome, inputs, constraints, definition of done), and the half about telling whether the answer is any good became a section of its own, Section 12, rather than the bottom of a long page people scrolled past. Teaching people to write prompts without teaching them to audit the output is the one thing in this area measured to make accuracy worse, so that material needed its own place in the contents. Added an unnumbered practice page after Section 12 that rebuilds the same brief four times with more taken away each time, short retrieval checks on most sections with an explanation for every option and no score anywhere, and rebuilt four comparisons to show the actual output each version produced rather than describing it.
ChatGPT Setup Guide v3.2, 8 August 2026
Re-verified every claim against OpenAI's own pages and cut anything we could not confirm. The model picker section was rebuilt around effort levels (Instant, Medium, High, Extra High), which replaced choosing a model by name. Moved a real win ahead of the privacy section and split privacy in two, so nobody has to read 630 words before getting something back. Promoted "it can be confidently wrong" to its own section. New sections on spotting the edge of what it knows, building a custom GPT (previously taught only behind the advanced wall, while three earlier sections assumed you knew it), Codex (now on every plan including Free), and what it costs to run including the shared agent credit pool. Added Scheduled Task caps by plan, per-plan Business pricing, and a personal and a business example to every section that lacked one.
Edition 2026.3
Earlier releases in this edition
Superseded
Claude Setup Guide v3.1, 19 July 2026
Added the "Where this stops working" section, re-based the design onto the Tempered Edge palette (graphite, bone, ember gold), unified every "Open the..." button onto one component, made wide tables scroll on their own so they cannot push the page sideways on a phone, and repointed two links that had started redirecting.
ChatGPT Setup Guide v3.1, 19 July 2026
Added the "Where this stops working" section, re-based the design onto the Tempered Edge palette, unified every "Open the..." button onto one component, made wide tables scroll on their own, and gave Section 12 the full Connectors to Apps to Plugins naming history with OpenAI's current help article cited.
Claude Setup Guide v3.0, 16 July 2026
Rebuilt the model lineup (Fable 5, Opus 4.8, Sonnet 5, Haiku 4.5), rewrote the privacy section for the training-on-by-default change, added a dedicated Claude Cowork agent section, refreshed plans and pricing, and expanded the advanced section. Every price, settings path, and link was checked against Anthropic's own pages on 16 July 2026.
ChatGPT Setup Guide v3.0, 16 July 2026
Re-checked every plan price, the model lineup, the settings paths, and every link against chatgpt.com/pricing. Renamed Connectors to Plugins to match the app, and removed claims that could not be confirmed against an official page.
Earlier versions
June to July 2026
Superseded
ChatGPT Setup Guide v2.0, July 2026
Rebuilt for the GPT-5.5 and GPT-5.6 model lineup, six pricing tiers with the new Go plan and ads, training-on-by-default privacy, ChatGPT Work and the unified desktop app, Scheduled Tasks, ChatGPT Sites, and the Plugin Directory.
Claude Setup Guide v2.0, June 2026
Added AI basics, account security, the file-safety decision tree, workflow recipes, and official courses.
Claude Setup Guide v1.0 and ChatGPT Setup Guide v1.0, June 2026
First releases. Model names, plans, and several features have changed since.
The most recent update went out on 6 September 2026. There is no fixed release calendar, and the terms do not promise one. What is promised is that every edition published lands in the same library buyers already use, at no extra cost, and gets listed here when it goes out. This page is the record of what is current.
Buy the current edition, get every edition after it.
One purchase covers both guides and every future edition for as long as I publish them. If a guide is ever retired, your access and the edition you bought stay yours.
$79 CAD plus tax, one time. 30-day satisfaction guarantee. A full refund ends the licence and guide access.