One-Prompt App Library 2.0

App: a Shopify product description generator that matches my brand voice and outputs four ready-to-paste lengths.
Model: Gemini 3.5 Flash.
Build in Google AI Studio Build mode as a client-side React app.
Data and context:
- Input fields: product name, 3 to 5 key features, target customer, price.
- A tone selector: casual, professional, luxury, playful.
- A "store voice" textarea where I paste two existing descriptions so output mirrors my voice.
Features:
1. One Generate button produces a 150 to 200 word storytelling description, 5 to 7 scannable bullet benefits, a 50-word short description for collection pages, and a 155-character meta description.
2. Each output block has its own copy button and a live word and character counter.
3. A Regenerate button per block so I can reroll one section.
4. A tighten and a lengthen control on the main description.
Validation:
- Enforce the meta description at 155 characters or fewer; red counter when over.
- Never invent specifications, materials, or claims absent from my inputs.
- If a required input is blank, prompt for it rather than guessing.
Deploy: AI Studio preview; run on my own free-tier key; paste outputs into Shopify by hand.
Acceptance: one product in, four on-brand, length-correct blocks out, ready to paste.

App: a Shopify SEO metadata generator with a live Google SERP preview and a social card preview.
Model: Gemini 3.5 Flash.
Build in Google AI Studio Build mode as a client-side React app.
Data and context:
- Inputs: page type (product, collection, page), name, a short description, the target keyword.
Features:
1. Generate 5 title-tag variations at 50 to 60 characters and 3 meta-description variations at 150 to 160 characters.
2. Generate a URL handle, Open Graph title and description, and Twitter card text.
3. Render a Google SERP preview card and a social preview card that update live as I edit.
4. Show a checklist score: keyword in title, keyword in description, optimal length, compelling call to action, each with a tick or cross.
5. Copy buttons on every field; a Copy all button.
Validation:
- Real-time character counters with color warnings; block titles over 60 and descriptions over 160.
- Keep the keyword in the first 50 characters of at least three title variations.
Deploy: AI Studio preview; own key; paste the chosen fields into the Shopify SEO section.
Acceptance: title and meta options that pass the checklist, with a preview of how they read in Google.

App: a Shopify product title optimizer that generates angles, scores them, and proposes A/B pairs.
Model: Gemini 3.5 Flash.
Build in Google AI Studio Build mode as a client-side React app.
Data and context:
- Inputs: current title, main keyword, product type.
Features:
1. Generate 10 title variations across five angles: SEO-first, benefit, audience, feature, emotional.
2. For each, show a 70-character counter and an SEO score from keyword placement, length, and readability.
3. Highlight which words are the keyword in each title.
4. Suggest two A/B test pairs and explain what each pair tests.
5. Copy individual titles or all of them.
Validation:
- Flag any title over 70 characters; the main keyword appears in at least half the variations.
- Do not duplicate the input title in the output set.
Deploy: AI Studio preview; own key.
Acceptance: ten scored titles and a clear A/B pair to test first.

App: an image alt-text generator that reads the actual image and writes SEO and accessible alt text.
Model: Gemini 3.5 Flash, using its multimodal image understanding.
Build in Google AI Studio Build mode as a client-side React app.
Data and context:
- Let me upload an image or paste an image URL.
- Inputs: product name and target keyword.
Features:
1. Analyze the image and produce a descriptive alt text and an SEO alt text containing the keyword, each under 125 characters.
2. Detect decorative images and flag them as needing empty alt text.
3. A batch mode that accepts several image URLs and returns a table of filename and alt text.
4. Export the table as CSV (filename, alt text).
5. Inline image thumbnails so I can see what each line describes.
Validation:
- Enforce the 125-character recommendation with a counter.
- Describe only what is visibly in the image; never assert an unseen brand, color, or detail.
Deploy: AI Studio preview; own key; paste alt text into Shopify image fields or import the CSV.
Acceptance: every product image leaves with accurate, keyworded, length-correct alt text.

App: a Shopify collection description writer producing hero and extended copy plus meta.
Model: Gemini 3.5 Flash.
Build in Google AI Studio Build mode as a client-side React app.
Data and context:
- Inputs: collection name, the products it contains, target keyword.
- A style selector: minimal, story-driven, feature-focused, lifestyle.
Features:
1. Generate a 50 to 100 word hero description for above the product grid.
2. Generate a 200 to 300 word extended description for below the grid, with who it is for and how to choose.
3. Generate a meta title and meta description.
4. Place the keyword naturally; never stuff it.
5. Copy buttons and character counters per block; a collection-page preview.
Validation:
- Keep the meta title at 60 characters or fewer.
- Confirm the keyword appears in the hero description once or twice, not more.
Deploy: AI Studio preview; own key.
Acceptance: hero copy, extended copy, and meta that read naturally and target the keyword.

App: a customer-review response writer that handles every rating in my brand voice.
Model: Gemini 3.5 Flash.
Build in Google AI Studio Build mode as a client-side React app.
Data and context:
- Paste a customer review; select a star rating from 1 to 5.
- For negative reviews, select an issue type: shipping, quality, sizing, service, wrong item.
- A store-name input and a tone note.
Features:
1. Generate a response that thanks the customer, addresses their specific points, offers a solution if negative, and invites future engagement.
2. Provide two alternative variations per review.
3. An editable field so I can adjust before copying.
4. A small template library for the most common situations.
Validation:
- The response must reference at least one detail the reviewer actually mentioned.
- Never use defensive or argumentative language, even for a one-star review.
Deploy: AI Studio preview; own key; paste into the review platform.
Acceptance: paste any review, get a calm, specific, on-brand reply in one click.

App: a product FAQ generator that also outputs valid FAQPage schema.
Model: Gemini 3.5 Flash.
Build in Google AI Studio Build mode as a client-side React app.
Data and context:
- Inputs: product name, description, key features, price.
Features:
1. Generate 10 to 15 FAQs covering specs, sizing, care, shipping, returns, comparisons, and durability.
2. Let me edit answers inline and reorder questions.
3. Export three ways: an HTML accordion to paste into Shopify; valid JSON-LD FAQPage schema; plain text.
4. Toggle individual FAQs on or off before export.
Validation:
- The exported JSON-LD must be valid JSON.
- The FAQPage Question count must equal the number of enabled FAQs on screen.
- Answers grounded in the inputs; do not promise policies I did not provide.
Deploy: AI Studio preview; own key; paste the accordion and the JSON-LD into the product page.
Acceptance: an editable FAQ set and matching schema that validates.

App: a product size-guide builder with conversions and an HTML export.
Model: Gemini 3.5 Flash.
Build in Google AI Studio Build mode as a client-side React app.
Data and context:
- Select product type: t-shirts, pants, dresses, shoes, hats, accessories.
- Paste my own size data or use a standard template.
- A fit descriptor: relaxed, regular, slim.
Features:
1. Generate a size-chart table (XS to 3XL or numeric).
2. Generate how-to-measure instructions and a between-sizes recommendation.
3. Generate US, UK, and EU conversions.
4. Export as an HTML table to paste into Shopify; show a mobile-friendly preview.
Validation:
- Confirm every size row has a value for every measurement column.
- Flag missing cells rather than inventing measurements.
Deploy: AI Studio preview; own key.
Acceptance: a complete, responsive size table and instructions ready to paste.

App: a Shopify product tag generator that proposes tags grouped by purpose.
Model: Gemini 3.5 Flash.
Build in Google AI Studio Build mode as a client-side React app.
Data and context:
- Inputs: product title, description, category.
Features:
1. Generate 30 or more tags grouped by type: category, attribute, style, occasion, audience, search-term.
2. Click tags to select or deselect, with a running count.
3. Copy selected tags comma-separated for Shopify.
4. Suggest a consistent naming convention.
Validation:
- Warn on any tag over 255 characters and block it from the copied set.
- Ensure each group has at least three tags.
Deploy: AI Studio preview; own key.
Acceptance: a curated, deduplicated tag set ready to paste into the product.

App: an email subject-line generator with preview text and a spam check.
Model: Gemini 3.5 Flash.
Build in Google AI Studio Build mode as a client-side React app.
Data and context:
- Inputs: email purpose, main offer, urgency level.
Features:
1. Generate 20 subject lines across styles: curiosity, urgency, personalization, question, list, emoji, benefit, FOMO.
2. Suggest matching preview text for each.
3. Show a character counter (50 recommended, 70 max) and a mobile truncation indicator.
4. Run a spam-word check and flag risky lines.
5. Offer two A/B test pairs; copy subject and preview together.
Validation:
- Flag subject lines over 70 characters and any that trip the spam-word list.
- Personalization variants include a clear name token placeholder.
Deploy: AI Studio preview; own key; paste into Klaviyo or Shopify Email.
Acceptance: twenty vetted subject lines and a first A/B pair to test.

App: a multi-platform social caption generator that adapts one message per network.
Model: Gemini 3.5 Flash.
Build in Google AI Studio Build mode as a client-side React app.
Data and context:
- Inputs: product or topic, key message, call to action, a brand-voice note.
Features:
1. Generate captions for Instagram (storytelling, up to 30 hashtags), TikTok (short hook), Facebook (conversational), Pinterest (searchable keywords), X (under 280 characters), and LinkedIn (professional), shown in tabs.
2. Regenerate per platform and edit inline.
3. Per-platform character counters and hashtag suggestions.
4. Copy button per platform.
Validation:
- Enforce each platform's character limit, especially X at 280.
- Keep hashtag counts within platform norms; never exceed 30 on Instagram.
Deploy: AI Studio preview; own key.
Acceptance: one input yields ready captions for six platforms, each within its limits.

App: a Shopify blog post generator that drafts in two stages, outline then full draft.
Model: Gemini 3.5 Flash.
Build in Google AI Studio Build mode as a client-side React app.
Data and context:
- Inputs: topic, target keyword, word-count goal, post type (how-to, listicle, comparison, buyer's guide).
Features:
1. First generate an SEO title, meta description, and an H2 and H3 outline.
2. On my approval, expand each section into a full draft.
3. Suggest internal-link anchors and a closing call to action.
4. Show a live word count and a keyword-density readout.
5. Export as HTML for Shopify.
Validation:
- Keep within 10 percent of the word-count goal.
- Confirm the keyword appears in the title, the first paragraph, and at least one H2.
Deploy: AI Studio preview; own key; paste the HTML into a Shopify blog post.
Acceptance: approve an outline, leave with a structured, keyworded HTML draft.

App: a catalog-wide SEO auditor that scores every product in one pass.
Model: Gemini 3.5 Flash.
Build in Google AI Studio Build mode as a full-stack app (React client, Node server).
Data and context:
- A file upload or large textarea where I paste my full Shopify product export as CSV (title, description, tags, handle).
- Use the one-million-token context window so the model reasons over every row in one pass; the window holds roughly 1,500 pages of text, so a full catalog fits.
Features:
1. For each product, score title length and keyword use, description length and quality, missing tags, and handle health.
2. Output a sortable table with a 0 to 100 score per product and the top three fixes for each.
3. Use an npm CSV parser the agent installs automatically.
4. Summary cards: average score, count of products under 50, most common gap.
5. Export the full audit as CSV.
Validation:
- The row count in the audit must equal the row count in the uploaded CSV.
- Flag any row that failed to parse rather than skipping it silently.
Deploy: keep in the AI Studio preview for personal use; the CSV stays in your session and is not shared.
Acceptance: paste a catalog, leave with a scored, sortable, exportable audit of every product.

App: a bulk product-description factory that rewrites a whole catalog in my voice.
Model: Gemini 3.5 Flash.
Build in Google AI Studio Build mode as a full-stack app.
Data and context:
- Upload a CSV of products (name, features, price).
- Paste my brand guide and two sample descriptions into a context panel; hold them in the one-million-token window so every generated description matches my voice and rules.
Features:
1. For every row, generate a 150-word description, five bullets, and a 155-character meta description.
2. Process in batches with a visible progress bar and a pause and resume control.
3. Use an npm CSV library to read input and write output.
4. Export a CSV ready for Shopify product import.
Validation:
- Output row count equals input row count.
- Each meta description is 155 characters or fewer.
- No banned word from my brand guide appears in any output.
Deploy: preview for personal use; or private Cloud Run if a teammate runs it on their own key. Keys are never placed in client code.
Acceptance: upload a catalog CSV, download a brand-consistent descriptions CSV ready to import.

App: a review-mining dashboard that analyzes a full review set at once.
Model: Gemini 3.5 Flash.
Build in Google AI Studio Build mode as a full-stack app.
Data and context:
- Paste hundreds of customer reviews as bulk text; use the one-million-token window to analyze them all together.
Features:
1. Extract an overall sentiment score, the most-mentioned positives and negatives, recurring themes, feature requests, and competitor mentions.
2. Render charts with the recharts npm package: a sentiment breakdown and a theme-frequency bar chart.
3. Pull quotable testimonials for marketing.
4. List specific product-description and FAQ improvements suggested by the reviews.
5. Export a summary report.
Validation:
- The sum of categorized reviews must equal the number pasted.
- Every extracted testimonial must be a verbatim quote present in the input.
Deploy: preview for personal use.
Acceptance: paste a review set, leave with charts, themes, testimonials, and a concrete improvement list.

App: a policy-aware customer-support reply composer.
Model: Gemini 3.5 Flash.
Build in Google AI Studio Build mode as a full-stack app.
Data and context:
- Paste my full shipping, return, refund, and warranty policies into a context panel held in the one-million-token window.
- Then paste a customer message.
Features:
1. Generate a reply that answers using only my actual policies, links to the relevant policy section, and offers clear next steps, in a chosen tone.
2. Show which policy passage each part of the answer is based on.
3. Offer two variations and an SMS-length version.
4. A template tray for the most common questions.
Validation:
- Every claim in the reply must trace to a sentence in the pasted policies.
- If the policies do not cover the question, say so and suggest escalation rather than inventing a policy.
Deploy: preview, or private Cloud Run for a support teammate. Keys stay server-side.
Acceptance: paste a message, get an on-policy reply with each claim traceable to my policy text.

App: a social content-calendar generator with CSV and iCal export.
Model: Gemini 3.5 Flash.
Build in Google AI Studio Build mode as a full-stack app.
Data and context:
- Inputs: month, posting frequency, and a list of upcoming launches, sales, and events.
Features:
1. Generate a full calendar with date, content type, topic, platform focus, best posting time, and the content pillar each post serves.
2. Balance content types automatically and include relevant holidays.
3. Render an interactive month grid; click a day for full details.
4. Use an npm date library the agent installs.
5. Export as CSV and as an iCal file.
Validation:
- The number of posts matches the chosen frequency across the month.
- No two adjacent days repeat the same content type.
Deploy: preview for personal use.
Acceptance: a balanced month of dated, typed posts I can export to my calendar.

App: a keyword cluster mapper that turns seeds into structured clusters.
Model: Gemini 3.1 Pro for the heavier reasoning.
Build in Google AI Studio Build mode as a full-stack app.
Data and context:
- Inputs: a seed keyword, product type, target audience.
- Optionally enable the Google Search tool for current trends.
Features:
1. Generate primary, long-tail, question, and commercial-intent keywords, grouped into topic clusters.
2. For each keyword show estimated difficulty, intent type, and the suggested page type (product, collection, or blog).
3. Render clusters as expandable groups.
4. Export as CSV.
Validation:
- Every keyword is assigned to exactly one cluster and one page type.
- Flag duplicates across clusters.
Deploy: preview for personal use.
Acceptance: a clustered keyword map with page-type assignments, exportable to CSV.

App: a competitor teardown tool that reads live pages and builds a matrix.
Model: Gemini 3.5 Flash with the URL Context tool enabled.
Build in Google AI Studio Build mode as a full-stack app.
Data and context:
- Inputs: up to three competitor URLs plus my own store URL.
- Use URL Context to read the live pages.
Features:
1. For each competitor, summarize product range and pricing, value proposition, target audience, messaging, strengths, and weaknesses.
2. Render a side-by-side comparison matrix and a gap analysis of what none of them offer.
3. List concrete differentiation opportunities for my store.
4. Export a report.
Validation:
- Every claim cites which URL it came from.
- Mark anything the pages did not reveal as not stated rather than guessing.
Deploy: preview for personal use.
Acceptance: a sourced comparison matrix and a gap list I can act on.

App: a pricing-strategy calculator that computes exact numbers and explains them.
Model: Gemini 3.5 Flash for the written guidance; compute math in code, not in the model.
Build in Google AI Studio Build mode as a full-stack app.
Data and context:
- Inputs: product cost, two or three competitor prices, a target profit margin.
Features:
1. Compute cost-plus, competitive, value-based, and psychological pricing.
2. For each strategy show the resulting price, margin percent, and break-even units, computed in JavaScript so the numbers are exact.
3. A discount-impact calculator: if I discount X percent, how many more units to hold profit.
4. Render a bar chart comparing strategies with recharts.
5. Export the analysis.
Validation:
- All financial figures are computed in code and must reconcile.
- The model writes only the explanation, never the numbers.
Deploy: preview for personal use.
Acceptance: exact prices and margins per strategy, with a clear recommendation and a discount calculator.

App: an ad-copy factory for Facebook and Google with strict, code-enforced limits.
Model: Gemini 3.5 Flash.
Build in Google AI Studio Build mode as a full-stack app.
Data and context:
- Inputs: product, main benefit, target audience, offer, and platform (Facebook or Google Ads).
Features:
1. For Facebook, generate primary text in three lengths, five headlines, and three descriptions.
2. For Google Responsive Search Ads, generate 15 headlines at 30 characters each and 4 descriptions at 90 characters each, with counters computed in code.
3. Show a SERP or ad preview.
4. Export an ad-set bundle as CSV.
Validation:
- Enforce every platform character limit in code and block any line that exceeds it.
- Run a prohibited-words check.
Deploy: preview for personal use.
Acceptance: platform-valid ad sets that never exceed a character limit, exportable to CSV.

App: an internal-link planner that maps links across my own pages.
Model: Gemini 3.5 Flash.
Build in Google AI Studio Build mode as a full-stack app.
Data and context:
- Paste the page or blog content I am working on.
- Paste a list of my other pages as title and URL pairs; hold the full list in context.
Features:
1. Suggest which pages to link to, the optimal anchor text for each, where in the content to place the link, and why it is relevant.
2. Prioritize suggestions by SEO value.
3. Identify orphan pages with no inbound links.
4. Export the complete linking plan as a table.
Validation:
- Every suggested link targets a URL from my pasted list; never invent a URL.
- Anchor text must appear plausibly in the content.
Deploy: preview for personal use.
Acceptance: a prioritized link plan that only references my real pages, plus an orphan list.

App: a brand-voice linter that checks any draft against my guide.
Model: Gemini 3.5 Flash.
Build in Google AI Studio Build mode as a full-stack app.
Data and context:
- Paste my brand guide once into a context panel held in the one-million-token window.
- Then paste any draft (product copy, email, post).
Features:
1. Check the draft against my guide for tone, banned words, sentence-length norms, and required phrases.
2. Output a pass or fail per rule with the exact offending text and a suggested rewrite.
3. Score the draft overall.
4. Keep the guide loaded so I can lint many drafts in a session.
Validation:
- Every flagged issue quotes the offending text from the draft.
- Never flag a rule the guide does not contain.
Deploy: preview, or private Cloud Run for a content teammate.
Acceptance: a per-rule pass or fail with quoted offenders and rewrites, reusable across drafts.

App: a customer-persona builder grounded in evidence when available.
Model: Gemini 3.1 Pro for richer reasoning.
Build in Google AI Studio Build mode as a full-stack app.
Data and context:
- Inputs: my products, price range, and what I already know about current customers.
- Optionally paste real review text to ground the personas in evidence.
Features:
1. Generate three to four distinct personas with demographics, values, pain points, goals, shopping behavior, objections, and the messaging that converts each.
2. Add marketing do's and don'ts per persona.
3. Render persona cards and a side-by-side compare view.
4. Export as a document.
Validation:
- If review text is provided, each persona trait must be supported by a pattern in that text.
- Label any trait that is an inference rather than evidence.
Deploy: preview for personal use.
Acceptance: three to four usable personas with conversion messaging, grounded where evidence exists.

App: a content operations hub for my team with saved history and sign-in.
Model: Gemini 3.5 Flash.
Build in Google AI Studio Build mode as a full-stack app.
Data and context:
- My brand guide stored once and reused for every draft.
Features:
1. Generate product copy, emails, and social posts from one workspace.
2. Save every draft, the content calendar, and version history to Firestore.
3. Sign team members in with Firebase Authentication using Sign in with Google.
4. Support multiplayer so two people can work without overwriting each other.
5. A searchable archive of past drafts by type and date.
Deploy: provision Firebase (Firestore and Auth) through the agent and deploy to Cloud Run.
Cost note: the deployed app uses your Gemini key for all users, so keep the key server-side and monitor usage.
Validation:
- Saved drafts persist across sessions; sign-in gates access.
- Two simultaneous editors do not overwrite each other.
Acceptance: a signed-in team workspace that drafts on brand and never loses work.

App: a customer gamification engine for my storefront with a persistent leaderboard.
Model: Gemini 3.5 Flash.
Build in Google AI Studio Build mode as a full-stack app.
Data and context:
- A spin-to-win or quiz game that issues discount codes, with weighted odds defined in code.
Features:
1. Collect an email before play; award weighted prizes.
2. Store scores and a leaderboard in Firestore.
3. Enforce session and daily play limits.
4. Provide a shareable result card and a Shop Now call to action.
Deploy: provision Firestore and deploy to Cloud Run, then embed on the storefront by link or iframe.
Cost warning, foregrounded: every shopper's Gemini calls and Cloud Run usage bill to your key. Keep keys server-side, cap calls per session, and watch spend before exposing this to traffic.
Validation:
- Prize odds match the configured weights over many plays.
- The leaderboard persists; the email gate blocks play until completed.
Acceptance: a working storefront game with a real leaderboard and a controlled, understood cost.

App: a brand brain that holds my whole brand and drafts on demand.
Model: Gemini 3.1 Pro for reasoning depth.
Build in Google AI Studio Build mode as a full-stack app.
Data and context:
- Persist my full brand guide, product catalog, and all policies in Firestore.
- Load them into the one-million-token window at run time so every output reflects the whole brand.
Features:
1. A single prompt box where I ask for any asset, a product description, an email, a policy-accurate answer, and get output grounded in the stored brand.
2. Save outputs to a library.
3. Let me update the stored brand documents through the app.
4. Sign in with Firebase Auth.
Deploy: provision Firestore and Auth and deploy to Cloud Run; keys server-side; monitor cost.
Validation:
- Outputs must not contradict the stored brand documents.
- A policy answer must trace to a stored policy; never assert an uncertified claim about the brand.
Acceptance: one box that produces on-brand, policy-accurate assets without re-pasting context.

App: an inventory planning board with persistent data and reorder math.
Model: Gemini 3.5 Flash for the written guidance; compute reorder math in code.
Build in Google AI Studio Build mode as a full-stack app.
Data and context:
- I enter or upload current stock, sell-through, and lead times per product and location; store it in Firestore.
Features:
1. Flag low-stock and overstock items.
2. Compute a suggested reorder quantity and date in code.
3. Keep multi-location notes.
4. Sign in with Firebase Auth so only my team sees it.
5. A dashboard with status badges.
Deploy: provision Firestore and Auth, deploy to Cloud Run, keys server-side; monitor cost.
Validation:
- Reorder quantities are computed in code and must reconcile to the inputs.
- The model writes only commentary, never the numbers; data persists across sessions.
Acceptance: a persistent board that flags reorders with exact, reconcilable quantities.

App: a review-to-roadmap tracker that turns recurring feedback into priorities.
Model: Gemini 3.5 Flash.
Build in Google AI Studio Build mode as a full-stack app.
Data and context:
- Paste new reviews on a recurring basis; store each batch in Firestore with its date.
Features:
1. Track theme and sentiment trends over time.
2. Surface rising complaints and feature requests.
3. Turn them into a prioritized improvement roadmap with a trend chart.
4. Sign in with Firebase Auth.
Deploy: provision Firestore and Auth and deploy to Cloud Run; keys server-side; monitor cost.
Validation:
- Every roadmap item links back to stored reviews that support it.
- Trend counts reconcile to the stored batches.
Acceptance: a living roadmap, evidenced by reviews, that updates as feedback accrues.

App: a UGC and testimonial library with tagging and safe reuse.
Model: Gemini 3.5 Flash.
Build in Google AI Studio Build mode as a full-stack app.
Data and context:
- I paste or upload customer testimonials and UGC references; store them in Firestore with tags.
Features:
1. Auto-tag each item by theme and sentiment.
2. Let me search and filter the library.
3. Generate ready-to-post captions from any item, with privacy-safe redaction of personal details.
4. Export selected items.
5. Gate access with Firebase Auth.
Deploy: provision Firestore and Auth, deploy to Cloud Run, keys server-side.
Validation:
- Generated captions quote the testimonial accurately.
- Redaction removes names and contact details when requested; the library persists.
Acceptance: a searchable proof library that turns any testimonial into a safe, on-brand post.

App: a product-launch command center with saved, reusable plans.
Model: Gemini 3.5 Flash.
Build in Google AI Studio Build mode as a full-stack app.
Data and context:
- Inputs: product, launch date, and features; store each launch plan in Firestore.
Features:
1. Generate a full sequence, teasers, launch emails, social posts, ad copy, on a timeline.
2. A task checklist and a countdown.
3. Save and reload past launches.
4. Sign in with Firebase Auth for the team.
Deploy: provision Firestore and Auth and deploy to Cloud Run; keys server-side; monitor cost.
Validation:
- Timeline dates are internally consistent relative to the launch date.
- Saved launches reload intact.
Acceptance: a reusable launch plan with copy, tasks, and a countdown the team can run.

App: an SEO audit tracker that stores history and shows trends.
Model: Gemini 3.5 Flash, optionally with the URL Context tool.
Build in Google AI Studio Build mode as a full-stack app.
Data and context:
- I audit a page (paste content or a URL) on a schedule; store each audit with its date and score in Firestore.
Features:
1. Score title, meta, headings, content, images, and links.
2. Store the score history per page.
3. Render a trend chart per page so I see improvement over time.
4. Sign in with Firebase Auth.
Deploy: provision Firestore and Auth and deploy to Cloud Run; keys server-side.
Validation:
- Each stored audit reconciles its line-item checks to its overall score.
- The trend chart reflects the stored history exactly.
Acceptance: a per-page score history with a trend that proves whether fixes worked.

App: a wholesale and B2B inquiry portal with sign-in and stored requests.
Model: Gemini 3.5 Flash.
Build in Google AI Studio Build mode as a full-stack app.
Data and context:
- A form where buyers submit company details and line-item requests; store submissions in Firestore.
Features:
1. Sign buyers in with Firebase Authentication.
2. Let them build a multi-line request with quantities.
3. Store each inquiry with a status.
4. Generate a clean summary email I can send back.
5. An owner view that lists and filters inquiries.
Deploy: provision Firestore and Auth, deploy to Cloud Run, keys server-side; monitor cost.
Validation:
- Submissions persist with correct line items and totals computed in code.
- Only signed-in buyers can submit; the owner view shows every stored inquiry.
Acceptance: a gated B2B intake that captures structured, status-tracked requests.

App: an email-sequence builder with persistence and ESP-ready export.
Model: Gemini 3.5 Flash.
Build in Google AI Studio Build mode as a full-stack app.
Data and context:
- Inputs: sequence type (welcome, abandoned cart, win-back, post-purchase) and brand voice; store sequences in Firestore.
Features:
1. Generate each email with subject, preview text, body, call to action, and send-timing trigger.
2. Save and reload sequences.
3. Export with Klaviyo and Shopify Email merge-tag placeholders.
4. Sign in with Firebase Auth.
Deploy: provision Firestore and Auth and deploy to Cloud Run; keys server-side.
Validation:
- Merge-tag placeholders match the chosen platform's syntax.
- Saved sequences reload intact; timings are ordered correctly.
Acceptance: reusable, ESP-ready email flows that persist and export cleanly.

App: a KPI dashboard fed by Google Sheets with a written summary.
Model: Gemini 3.5 Flash.
Build in Google AI Studio Build mode as a full-stack app.
Data and context:
- Connect a Google Sheet of my store metrics using AI Studio's Workspace integration.
- Optionally store snapshots in Firestore.
Features:
1. Render KPI cards, revenue, orders, conversion, average order value, from the Sheet.
2. Render trend charts and refresh on load.
3. Write a short plain-language summary of what changed.
4. Sign in with Firebase Auth so only my team sees it.
Deploy: provision the Workspace connection and Firebase, deploy to Cloud Run, keys server-side; monitor cost.
Validation:
- Every figure on the dashboard matches the source Sheet.
- The written summary follows only from the displayed numbers.
Acceptance: a signed-in dashboard whose numbers match the Sheet, with a plain-language readout.

App: an affiliate and referral tracker with per-affiliate access.
Model: Gemini 3.5 Flash for written copy; compute totals in code.
Build in Google AI Studio Build mode as a full-stack app.
Data and context:
- Store referral codes, referrers, and recorded conversions in Firestore.
Features:
1. Issue and store unique codes; record conversions against each.
2. Compute payouts in code; show a leaderboard.
3. Generate referral invite copy on demand.
4. Sign affiliates in with Firebase Auth so each sees only their own stats; an owner view sees all.
Deploy: provision Firestore and Auth, deploy to Cloud Run, keys server-side; monitor cost.
Validation:
- Payout math is computed in code and reconciles to recorded conversions.
- Access control isolates each affiliate's data; codes are unique.
Acceptance: a gated tracker that issues unique codes and reconciles payouts to conversions.