by VariatCopy (LX AI)
Open VariatCopy, provide the required inputs, run Demo to preview the workflow, then switch off Demo when a live key is configured.
Two angles for every product, ready to test.
Decision-support only — confirm before you publish or act.
When you first open the app you see a blank project card. Click “New project”, give it a short name that matches the product you are working on, and press Enter. The app creates a folder .data/ inside the project directory where it stores demo runs and any saved configurations.
Each project needs three pieces of information: product name, core benefit, and target audience. Fill these fields in the form at the top of the screen. The product name is used only for labeling the output; the core benefit should be a single phrase that describes what the product does for the user; the target audience should be a brief description of the person who makes the buying decision.
Press the “Demo” button. The app uses a built‑in mock model to generate two contrasting copy angles instantly. You will see two cards appear below the form, each containing a hook, a body paragraph, a CTA, and a one‑line test hypothesis. Use this preview to verify that the inputs are being interpreted as you expect.
When you are satisfied with the Demo preview, locate the API key field in the settings pane. Paste a valid key from your chosen LLM provider. Toggle the Demo switch off. The app will now call the live model each time you press “Generate”.
With the live key active, click “Generate”. The app sends a prompt that instructs the model to produce two opposite angles for the same product. One angle emphasizes a discount or price‑based incentive; the other emphasizes quality, durability, or prestige. The model returns JSON that the app parses into the UI.
Each variant card shows:
Read the hook and body aloud. If any phrase feels generic, edit the text directly in the card. The app saves your edits locally under .data/edits.json.
Press the “Save” button on each card to lock the current version. Saved variants appear in the “History” panel on the left, timestamped and labeled with the product name.
The hypothesis should reference a specific metric you can track after publishing (click‑through rate, conversion rate, time on page, etc.) and a direction of change (increase or decrease). Keep it under 20 words so it remains actionable.
When you save a variant, the app automatically copies the hypothesis into a markdown file hypotheses.md inside the project folder. Each line follows the format:
[YYYY-MM-DD] Variant A:
You can later diff this file against your analytics export to see which predictions held true.
After a test run, open hypotheses.md. If the outcome matches the prediction, note the variant as a winner. If not, examine the hook and body for possible mismatches with audience expectations, then adjust the core benefit field and generate a new pair.
When you expose generated copy to users, treat the output as user‑generated content. The https://owasp.org/www-project-top-10-for-large-language-model-applications document outlines prompt injection risks and recommends sandboxing the model calls. VariatCopy runs the model in an isolated process and does not store prompts longer than the current session.
For broader AI risk management, refer to the https://www.nist.gov/itl/ai-risk-management-framework framework. It suggests documenting data provenance, monitoring for bias, and establishing human‑in‑the‑loop review steps—all of which map to the decision‑support disclaimer in the app.
All inputs you type remain in the browser’s memory until you press “Generate”. The request payload is sent over HTTPS to the model endpoint you configured. No prompt or output is written to external servers by VariatCopy itself. Demo runs are kept only in .data/demo/ and are cleared when you close the browser tab.
In the settings pane you can slide the “Temperature” control from 0.0 to 1.0. Lower values produce more deterministic phrasing, which is useful when you need strict adherence to a brand voice. Higher values increase variability, helping you explore unexpected angles.
The “Max tokens” field caps the length of each generated section. Set it to 60 for hooks, 120 for bodies, and 20 for CTAs if you want to enforce strict length limits.
If you have a specific formatting requirement (for example, you need the hook in uppercase), you can edit the prompt template stored at .data/prompt.txt. The file uses Handlebars‑style placeholders: {{productName}}, {{coreBenefit}}, {{audience}}. Save your changes and reload the app to see them take effect.
I built VariatCopy to give you a fast first draft, not a guaranteed publishing solution. The app does not:
Treat every output as a suggestion. Review it for brand fit, regulatory compliance, and factual accuracy before you use it in any campaign.
I keep a written checklist next to every release that touches this topic. The list is boring on purpose: name the owner, name the evidence artifact, and name the rollback. If a step cannot produce a file, a ticket, or a log line, it is not a step — it is a wish.
| Signal | What to record | Stop-ship? |
|---|---|---|
| Unexpected write or send | Tool name, args, identity | Yes until gated |
| Missing confirmation on destructive action | Call path and bypass | Yes |
| Untrusted text treated as instruction | Source surface and resulting action | Yes for agent reach |
| Scanner clean, judgment open | Deferred items + owner | Only if core path safe |
variatcopy helps with the measurable first pass. Architecture choices, legal sign-off, and customer-facing claims still need a human owner. Treat every checklist row as decision support, not a certificate.
When a prospect asks whether the product "covers everything," I answer with the boundary: automated checks catch known patterns; novel chains and policy judgment still need review. That honesty keeps support load down and keeps the blog aligned with what the product actually does.
If you need a second opinion, pair this write-up with the references below and with your own staging reproduction. Do not skip the reproduction step just because a model wrote a confident summary.
Two angles for every product, ready to test.
Operators and builders who need a fast first draft or checklist from VariatCopy.
Yes in explicit Demo mode. Live AI requires a configured key.
No. Outputs are decision-support; you still review before publishing or acting.
Only if those features are implemented in this product build — do not assume them from marketing copy.
Runs may be stored locally under the product's .data/ boundary; treat demos as ephemeral.