Skip to main content

AI Text Post Generator

The AI Text Post Generator is a browser-based writing tool that turns a short prompt into a finished text post using an open-source language model that runs locally on your device. It loads the distilgpt2 model through @xenova/transformers and produces a draft without sending your prompt to a paid API. When the model cannot load on a given device, the generator falls back to a structured, template-based draft so you always get usable output. There is no API key and no cost.

What it does​

The generator takes your topic or idea and shapes it into a post that fits where you want to publish. You describe the subject, choose how it should sound and how long it should be, and the tool assembles a draft you can copy, refine, or send onward. Because the model runs client-side, generation happens on your machine rather than on a remote server. This keeps the workflow aligned with the zero-cost model that the whole app is built around.

The output is plain text intended for social posts, captions, and short-form updates. You can regenerate as many times as you like, since each run uses your own device rather than a metered service.

The options and controls​

You can shape every draft with the following controls:

  • Tone (7 options): professional, casual, friendly, humorous, inspirational, urgent, and informative. The tone guides word choice and rhythm so the same topic can read as a calm explainer or an energetic announcement.
  • Length (3 options): short, medium, or long. This sets the rough size of the draft so it matches the format you are writing for.
  • Sentiment (3 options): positive, neutral, or negative. Sentiment steers the emotional leaning of the copy.
  • Language (12 options): generate posts in any of twelve supported languages.
  • Target platform: select the platform you are writing for, and the generator shapes the length and format to suit it.
  • Hashtag suggestions (optional): turn this on to append suggested hashtags to the draft.
  • Emoji suggestions (optional): turn this on to include suggested emoji where they fit.

You can also apply a fine-tuned LoRA adapter to nudge the output toward a consistent brand voice. See fine-tuning for how a brand-voice adapter is trained and applied.

How it works: client-side with a fallback​

When you generate a post, the tool first tries to run distilgpt2 in your browser through @xenova/transformers. The model weights load into the page and inference happens on your own hardware. If that model is unavailable — for example on a constrained device or a browser that cannot allocate the needed resources — the generator switches to a structured template-based draft built from your selected tone, length, sentiment, language, and platform. The template path is deterministic and dependable: it never silently fails, and it keeps the experience working everywhere.

This two-path design is the same honesty-first approach used across the app. The browser-based model is genuine open-source AI, and the template fallback is a transparent backup rather than a hidden limitation. For the full picture of how local inference is wired, read how the AI works.

Where results go​

Every generated post is saved to your "My Posts" history page, so you can return to a draft, compare versions, or reuse copy later. From there you can move a post toward distribution — for example by lining it up in the scheduling workflow when you are ready to publish.

The text generator is one of several creators in the app. If you also need visuals or other formats, see the tools overview and the sibling creators for images, video and reels, audio and music, 3D models, and surveys and polls. New to the app? Start with the introduction and the quick-start guide.

Frequently asked questions​

Do I need an API key or a paid account to generate posts? No. The generator runs the open-source distilgpt2 model in your browser and falls back to a template-based draft when the model cannot load. There is no API key, no subscription, and no per-use charge.

What happens if the model fails to load on my device? The tool automatically switches to a structured, template-based draft that uses your selected tone, length, sentiment, language, and platform. You still get a complete post; it is simply assembled from a template rather than the model.

Can I keep a consistent brand voice across posts? Yes. You can apply a fine-tuned LoRA adapter to steer the output toward your brand voice. See the fine-tuning page for details.

Which languages are supported? The generator supports twelve languages, selectable per draft.

Where do my generated posts end up? Each post is saved to your "My Posts" history page, where you can revisit, refine, and reuse it, including handing it off to the scheduling workflow.