How to read this log. Entries are added only after a change is live in buyer kits — every date below is a ship date. This is a record, not a roadmap. (Each kit is an online hub — your private workspace — plus a PDF companion for offline use, in four platform editions: ChatGPT, Claude, Copilot and Gemini.)
The homepage now leads with the Capital PM Toolkit
The homepage now opens with the Capital PM kit: the hero example is the bid-leveling prompt, and the lead copy speaks to change orders, vendor bids and EVM reports run through your own AI. The Legal kit and the confidentiality story keep their place directly below. Kit contents and pricing are unchanged.
Optional blanks and lists: the fill surface now says what happens
Two clarity fixes from our first round of beta feedback. When you leave an optional detail blank, the copied prompt keeps a visible [placeholder] and your assistant treats it as not stated; the hub now says so up front instead of leaving you to wonder. And prompts that accept lists now tell you that dashes or numbered lines carry into the prompt word for word. Four fields that expect multi-line input (project scope, conflict-check lists) also got proper multi-line boxes.
Easier tapping on phones, plus small storefront polish
The fill-mode toggle (Side-by-side / Tightened / Guided) now meets the 44px touch-target standard on phones, matching the rest of the hub’s mobile controls. On the storefront, the Legal kit page’s buy box got a copy cleanup and the homepage’s main button now uses the deeper coral that passes accessibility contrast at rest, not just on hover.
The Invention Test — public receipts for the anti-invention rails
We published a live, dated test: the same deliberately incomplete pipeline through a naive prompt (10 of 10 runs invented dollar figures) and through the kit’s forecast prompt (10 of 10 clean, “[NOT PROVIDED]” where data was missing). Unedited transcripts included; re-run monthly.
Kit pages re-checked against the shipped kits
A parity audit covered all five kit pages against the shipped kits; three needed updates. The PR page now lists the current nine work tasks (including Spokesperson Briefing & Tough-Q&A Prep) and its sample prompt is the kit’s current hardened version, with the [VERIFY] guardrail shown. The Accounting page lists all nine work tasks — client communications included — plus the numbers & anti-invention reference card. The Sales & RevOps page now states the numbers rule its pipeline and forecast prompts enforce: every dollar figure in the output is one you pasted, or it comes back “[NOT PROVIDED]”.
Purchase history and receipts on your account page
Each kit on your My Account page now shows when you bought it, what you paid, and a link to view your receipt. The page also gained a direct support line (support@sharpwork.ai, with the 14-day refund window stated) and sign-in copy that covers invited users as well as buyers.
PDF companions now mirror the kit hubs one-for-one
Every kit’s PDF companion was restructured so modules and prompts match the online hub exactly: same module order, same prompt titles, same hardened prompt text (Legal 8 modules/31 prompts, Sales 8/26, Accounting 10/27, PR 9/22, PM 14/45 — PM’s re-issue shipped earlier the same day with the rebuild below). Every PDF now also includes the Single-User License and the 14-day money-back guarantee in writing.
Stronger number-safety rails on money-handling prompts
The PM scope-change pricing prompt gained explicit pricing-basis fields plus a NUMBERS RULE — every figure must come verbatim from your inputs or shown arithmetic, and anything missing is marked ‘Not stated — estimate with your team’ instead of guessed. Sales reporting and ROI prompts and the Accounting tie-out prompt got matching SELF-CHECK rails, in both the hub and the PDF.
Attach the file instead of pasting it
Work tasks whose real input is a document — risk briefs, scopes, vendor bids, meeting transcripts, invoices — now tell you to attach the file directly to your own AI (ChatGPT, Claude, Copilot and Gemini all read Word and PDF natively), with the paste box kept as a labeled fallback. Document fields show an “Attach it in ChatGPT” badge, worded for your edition’s platform — the file goes to your AI, never to Sharpwork.
PM kit rebuilt: 14 work tasks for scope, schedule and budget
The Project Management kit was rebuilt around the triple constraint — Scope, Schedule, Budget — growing from 9 to 14 work tasks, with new modules for Business Case, RFQ/RFP, Bid Leveling, SOW, Change Orders, Closeout, Schedule Baseline & Critical Path, Cost Estimate & Budget Development, and EVM performance reporting. The certification question bank was rebuilt to match (39 questions grounded in the new modules) and the PDF companion re-issued with the same 14 workflows in the same order. If you bought the PM kit before this date, your hub already shows the rebuilt kit — updates land automatically, and your original download link serves the current PDF.
Paste a spreadsheet — or just upload the file
When you paste multi-line data (a P&L, a pipeline export, a CSV) into a prompt field, the assembled prompt now wraps it in a fenced literal data block, so your AI reads it as data rather than instructions and field boundaries stay crisp. Every paste field also gained an “Upload a file” button with drag-and-drop — the file is read entirely in your browser and never sent to us.
Guardrails against invented numbers across the prompt library
Ten high-stakes prompts were hardened: NUMBERS RULE preambles (money figures come only from what you provide), SELF-CHECK blocks so the AI audits its own draft for invented figures before you see it, and [VERIFY CURRENT] markers on every form number, statute, threshold and deadline. The same hardening was ported into 16 buyer PDFs the same day, so the PDF editions of those prompts carry the same rails.
New free pages: the confidentiality method, spelled out
Published sharpwork.ai/security/ — a full walk-through of the [Client] bracket anonymization method the kits are built on, including an honest-limits section — plus a 10-question FAQ page.
Copy-to-AI buttons now pre-fill the prompt
After you copy a prompt, the “Open ChatGPT” / “Open Claude” link now carries the blank template straight into the composer, so it lands pre-filled for review instead of opening an empty tab. Filled prompts — the ones containing your own client data — stay clipboard-only, so your data never rides a URL or browser history.
Three ways to fill in a prompt
Filling in a prompt now offers three modes — Guided (one question at a time), Side-by-side (your fields next to a live preview of the finished prompt), and Tightened (every field in one compact list) — with protected placeholders like [Client] shown as clearly marked chips so you can see exactly what stays anonymized. The copy button names your platform (“Copy & paste into Claude”), and every task opens in a focused, single-task view.
Easier scanning, bigger tap targets
Prompt Library cards were re-ranked for quick scanning — larger titles, a small category label, and a quieter sample preview — and every small control got a proper 40px tap target with press feedback. Headings were re-wrapped to break in cleaner places.
See each kit’s scope and stakes at a glance
The Core Playbook now opens with a scope manifest (work tasks, ready prompts, worked examples, PDF companion, certification — real counts), every work-task row shows its prompt count, estimated time saved and a risk marker, and a once-per-visit “Where to start” tip points you at your kit’s highest-value task. Mobile reading was flattened into a cleaner single-surface layout, and the Prompt Library gained its own manifest.
Kit hubs now open with their highest-stakes work
Each hub now opens with its highest-stakes value instead of setup logistics: protection for Legal and Accounting, reputation for PR, effectiveness for Sales, clarity for PM. The Start Here steps are all visible at once (including the “review the result yourself” step), a “Where to start” chooser deep-links to each kit’s top tasks, and “Highest value” markers flag the flagship rows.
Free prompt pages for every profession
Five free, no-signup prompt pages went live — one per vertical — each with three complete copy-paste prompts taken verbatim from the paid kit (the three highest-intent tasks, e.g. Legal’s billing narratives and client-intake memo), a worked before-and-after example, and the methodology behind them.
Your account, one click from every kit hub
The kit hub’s top rail now shows your signed-in state and links straight to My Account, and every “recover access” pathway in the hubs now points to the account front door instead. The signed-in check reads only your own browser cookie — nothing else is exposed.
Core Playbook prompt cards redesigned, 130 titles rewritten
Prompt blocks moved from a dark terminal look to a light document style, fill-in fields now come first, and all 130 task and prompt titles across the five kits were rewritten to lead with the outcome (“End-of-Day Billing Batch” became “Clean up today’s billing”). A dismissible first-run helper orients new buyers, and touch targets were enlarged with clearly visible focus states.
My Account: passwordless sign-in to everything you own
My Account is live — enter your purchase email, get a magic link, and see every kit you own in one place with one-click open into each hub. A “Sign in” entry was added to the storefront navigation.
Fuller safe-use setup guidance in the ChatGPT editions
The ChatGPT editions of the PM, PR, Accounting and Sales kits gained the full data-protection setup note — training opt-out guidance plus Team and Enterprise workspace notes — the same depth carried by the Claude, Copilot and Gemini editions.
Twelve months of updates are included in the one-time price — everything above shipped under that promise. Questions about any of these updates: hello@sharpwork.ai.