Textory — AI rewriting of product descriptions

Textory is LIONEX's own SaaS product. The platform automatically rewrites the product descriptions of an online store into unique texts structured for SEO. We do not only build sites for clients — we also build and keep in production our own product, with processing queues, an AI integration, billing and a public API.

The problem it solves

Stores upload descriptions from the supplier — the very same texts sit in the catalogues of dozens of competitors. Rewriting a few thousand items by hand takes months. Textory reduces any catalogue to a single product model, rewrites the descriptions through AI and returns a ready feed in the format of your platform.

A unique description removes a technical loss — duplicates, thin content, the absence of structure. On its own it does not create demand and does not guarantee rankings: ranking is affected by assortment, prices, links and user behaviour.

How it works

1. Import. A link to a feed or a file — XML, YML, CSV, Excel. Sources: Prom.ua, Horoshop, Shop-Express, OpenCart, WooCommerce.
2. AI rewriting. A queue processes the products; each one takes a single call to Google Gemini, which returns the description and, if needed, a new title, keywords and a meta description. Progress is visible in real time and processing can be paused.
3. Export. The finished catalogue comes out in the format of the platform (Prom.ua, Rozetka, XML, CSV, JSON), with a choice of which fields to output. The file is ready to import without manual edits.

What is inside

• Bulk rewriting that preserves the product specifications.
• SEO structure: H2/H3 headings, specification lists.
• Generation of keywords and a meta description in a single call.
• The name_ua / description_ua fields for Prom.ua.
• GEO mode — optimising a description for AI search (ChatGPT, Perplexity, Google AI Overviews): a direct factual answer, question subheadings, fact density — plus a deterministic GEO score from 0 to 100.
• Settings for style (sales / neutral / technical), length and emphasis; a live preview of a single product.
• Integrations: REST API, webhooks, API keys, public feeds, a module for OpenCart.
• Data protection: an NDA mode; catalogues are stored encrypted; the data is not passed to third parties and does not go into AI training.

GEO mode is a mechanism: structure and fact density that AI systems find easier to cite. Nobody — ourselves included — can promise a mention in an AI answer.

Technically

The frontend is Next.js + React. The backend is Fastify + PostgreSQL + Prisma, queues on BullMQ + Redis, AI on Google Gemini. Deployment through Coolify. In an external measurement on 02.08.2026 the Textory page showed a TTFB of 0.22 s at 141 KB of HTML — the fastest result among the 16 sites we measured; the page carries structured data (SoftwareApplication, Organization, FAQPage) and a PWA manifest. This is the state of the site on the date of the measurement.

How it differs

Unlike generic AI writers (ChatGPT, Jasper, Copy.ai), which generate text in a chat, Textory works with a product feed: it takes a catalogue from the platform and returns an import-ready feed for Prom and Rozetka. Unlike manual copywriting, it scales that to thousands of items in a queue with pause and resume. Unlike importing supplier descriptions as they are, it rewrites them into unique ones. Plus the Ukrainian specifics (name_ua/description_ua), generation of keywords and meta in a single call, a measurable GEO score and transparent payment per character.

The comparison is at the level of capabilities, without disparagement. Textory does not replace strategy and assortment and does not guarantee rankings: it removes the technical loss from duplicates and thin content.

Try it

Textory is a self-service product with transparent payment for the volume of text. Demo catalogues for testing hold up to 11 806 products. If you need custom catalogue automation or an integration for your platform, leave the task with the agency: we reply within 2 hours and work under a contract.