How to use AdCopy Gen

Setting up your workspace

Open the application

I start by launching AdCopy Gen from the desktop shortcut or the web portal. The login screen asks for a username and password that I created during the onboarding email. After I sign in, the landing page shows a blank canvas with a top bar that reads “New Project”. I click that button to create a fresh workspace for each campaign I plan to run.

Create a project folder

In the dialog that appears I type a short, descriptive name for the project, such as “Spring‑Sale‑Google‑Ads”. I then choose a location inside the product’s data directory; the tool automatically creates a subfolder named after the project under .data/. Inside this folder the app stores demo runs, temporary JSON snippets, and any configuration files I later add. I verify the folder appears in the file explorer pane on the left side of the UI.

Adjust basic preferences

Before I add any content I open the Settings gear icon. I set the default language to English, choose the tone preset “Neutral” as a starting point, and turn on the autosave toggle so that every change is written to disk after two seconds of inactivity. I also enable the “Show token count” option because it helps me stay within the model’s context limit when I later paste long product descriptions.

Defining product and goal inputs

Product description fields

The main form contains three large text boxes. In the first I paste a concise product summary that includes the brand name, the core function, and the primary differentiator. For example, I might write: “Eco‑Fit‑Tracker, a wrist‑worn device that measures steps, heart rate, and sleep quality, made from recycled aluminum and offering a seven‑day battery life.” I keep this under 300 characters to avoid truncation.

In the second box I list the target audience segments. I write each segment on a new line: “Busy professionals aged 25‑40”, “Parents looking for kid‑friendly activity tracking”, “Fitness enthusiasts who want detailed VO2‑max estimates”. This helps the model generate copy that speaks directly to each group.

The third box is for any mandatory legal or brand‑specific wording that must appear in every ad, such as a trademark symbol or a disclaimer like “Results may vary”. I copy‑paste the exact phrasing from the brand guidelines.

Goal selection

Below the text boxes sits a dropdown labeled “Campaign goal”. I choose one of the preset options: “Click‑through”, “Lead generation”, “Brand awareness”, or “Product launch”. Each goal changes the internal weighting the model uses when it crafts hooks, body copy, and calls‑to‑action. For a lead‑generation goal I expect the model to emphasize benefit statements and a strong CTA like “Sign up for a free trial”.

Saving the input set

After I fill in all fields I click the “Save as template” button. The tool writes a JSON file named inputs.json into the project folder. I can reuse this template for future variations by loading it from the “Load template” menu, which saves me from retyping the same product details.

Running Demo mode to preview the workflow

What Demo mode does

Demo mode runs the same generation pipeline but substitutes a deterministic, rule‑based placeholder for the live language model. This lets me see the shape of the output without consuming any API credits or sending data to an external endpoint. The placeholder inserts token‑level markers such as [HOOK], [BODY], and [CTA] that I can later replace with real text.

Starting a Demo run

With the inputs saved I press the large “Run Demo” button at the bottom of the form. A spinner appears for two seconds, then the UI expands to show three columns labeled “Google”, “Meta”, and “LinkedIn”. Each column contains a card with the placeholder sections filled in.

Reading the preview output

I examine each card to verify that the structure matches my expectations. The hook should be a short, attention‑grabbing phrase (e.g., “Track your impact, not just your steps”). The body should expand on the benefit and include any mandatory disclaimer I supplied. The CTA should reflect the goal I chose, such as “Learn more” for brand awareness or “Get started free” for lead generation. I note any places where the placeholder text feels off‑topic; those are the spots I will focus on when I switch to live mode.

Saving the Demo preview

There is a “Download preview” link that exports the three cards as a markdown file. I store this in the project folder under demo_preview.md so I can compare it later with the live output and measure how much the model changed the tone or wording.

Switching to live mode with an API key

Where to add the key

Live mode requires a valid API key from the model provider. I open the Settings panel again, navigate to the “API credentials” tab, and paste the key into the field labeled “Provider API key”. The field masks the characters as I type, and a small check‑mark appears when the format looks correct (starts with sk- for the provider I use). I then click “Validate”. The tool makes a lightweight ping to the provider’s endpoint; if the response returns a 200 status I see a green banner that reads “Key active”.

Safety checks before enabling live

Before I flip the switch I re-read the ad-policy basics for the platform I am targeting. Google Ads, Meta, and LinkedIn treat claims differently, and the usual trip hazards are health claims, financial performance claims, and before-and-after imagery. I keep a short list of the claims I intend to make so I can substantiate each one before the campaign spends money.

Turning off Demo and enabling live

With the key validated I return to the main form. The Demo toggle switch, which was previously set to “On”, now shows a label “Live mode”. I click it to turn Demo off. The button text changes from “Run Demo” to “Run Live”. I also see a warning banner that reminds me that live calls will consume credits and that outputs should be reviewed before publishing.

First live run

I press “Run Live”. The spinner appears again, this time for a slightly longer period as the model processes the request. When the results load I see the same three‑column layout, but now the placeholder markers have been replaced with actual sentences. I immediately compare each variant to the Demo preview I saved earlier, noting differences in word choice, tone, and length.

Generating and refining ad copy variants

Generating the first set

The live run returns three variants per platform, each with a distinct hook angle. For Google Ads I might see: Hook: “Cut your carbon footprint while you train”, Body: “Eco‑Fit‑Tracker logs every step, heartbeat, and night‑rest using sensors built from reclaimed metal. The seven‑day battery means you never miss a workout, and the companion app gives you actionable insights to improve your routine.” CTA: “Shop now”. I copy each variant into a separate Google Doc for side‑by‑side comparison.

Using tone presets

If the first set feels too formal I open the tone preset dropdown and select “Friendly”. I rerun the live generation with the same inputs; the model adjusts the vocabulary, adding contractions and more colloquial phrasing. For example, the hook becomes “Stay green, stay strong”. I repeat this process with the “Bold” and “Concise” presets to build a small library of tonal options.

Creating A/B angle options

To test different messaging angles I manually edit the product description field before a new run. I create two copies of the base inputs: one that emphasizes durability (“Built to survive rain, sweat, and intense workouts”) and another that highlights health insights (“Get personalized heart‑rate zones and sleep scores”). I save each as inputs_durability.json and inputs_health.json. Running live mode on each file yields two distinct families of copy that I can allocate to separate ad sets.

Iterative refinement loop

After I collect the variants I follow a three‑step loop: - Read each copy aloud to spot awkward phrasing. - Mark any sentences that exceed the platform’s character limit (Google Ads headlines ≤30 characters, description ≤90 characters). - Edit the offending lines directly in the Doc, then feed the revised text back into the tool as a “seed” by pasting it into the product description box and running a new live generation with the temperature set low (0.2) to encourage the model to stay close to my edits. I repeat the loop until every variant meets the platform limits and reads naturally.

Exporting the final set

When I am satisfied I select all variants in the UI and click “Export bundle”. The tool creates a ZIP file named adcopy_bundle_.zip containing three subfolders (Google, Meta, LinkedIn). Inside each folder are plain‑text files named variant_1.txt, , . I upload these files to the ad‑platform’s bulk‑edit sheet or copy‑paste them into the ad creator UI.

What this is NOT

No guarantee of performance

I do not promise that any piece of copy generated by AdCopy Gen will achieve a specific click‑through rate, conversion volume, or return on ad spend. The model produces text based on patterns in its training data; actual results depend on many factors outside the tool’s control, such as audience targeting, bid strategy, and creative fatigue.

Not a replacement for human review

The output is decision‑support only. I must read every line, check for brand‑voice consistency, verify that no prohibited claims appear, and confirm that any required legal wording is present before I push the ad live. Skipping this step risks publishing copy that violates platform policies or misrepresents the product.

Does not automate publishing or budget management

AdCopy Gen stops at copy creation. It does not connect to ad accounts, adjust bids, pause underperforming ads, or allocate budget. Those actions remain in my hands or in the separate campaign‑management tools I use.

Does not retain or share your data beyond the local boundary

Demo runs store temporary JSON in the project’s .data/ folder and are cleared when I close the project or manually delete the folder. Live calls send only the product description and goal to the model provider; the tool does not keep a copy of the request or response after the generation completes. I still review the provider’s privacy policy to understand how they handle the data I transmit.

References

- Google Ads policies - Meta Advertising Standards - FTC truth-in-advertising guidance

Frequently asked questions

What is AdCopy Gen?

Ad copy that sounds human, not templated

Who should use AdCopy Gen?

Operators and builders who need a fast first draft or checklist from AdCopy Gen.

Does it work without an API key?

Yes in explicit Demo mode. Live AI requires a configured key.

Does it guarantee outcomes?

No. Outputs are decision-support; you still review before publishing or acting.

Does it include SSO, Slack, or bulk CSV?

Only if those features are implemented in this product build — do not assume them from marketing copy.

Where does data go?

Runs may be stored locally under the product's .data/ boundary; treat demos as ephemeral.

References

- schema.org — FAQPage - Google Search Central — Structured data

Try AdCopy Gen

Related reads

- AdCopy Gen overview — AdCopy Gen - The LX AI Micro-SaaS Topic Matrix