Logo
ХелпдескУправление знаниямиБизнесуТарифыЧто такое KCS?Блог
Русский
Личный кабинет
Работа с ИИ
© 2023 - 2026 ООО «Свормика». Swarmica - Платформа управления знаниями и ресурсами техподдержки. Версия: 6.1.2

#2148: Using AI agents for SEO/GEO/AEO optimization of the knowledge base

Отредактирована: сегодня

Starting from v6.1.0, you can create an AI agent with the SEO/GEO/AEO optimization competency in Swarmica. The agent automatically fills in the SEO fields and JSON-LD markup of published articles so that search engines and AI services find and rank the knowledge base correctly.

What the AI agent does

The agent works with published articles. For each language of an article that has a title and text, it fills in the empty fields:

  • seo_title — the title for search results and previews;
  • seo_description — a description of up to 160 characters;
  • seo_url — the relative address of the article;
  • seo_image — the first image uploaded to the article (if there is one);
  • seo_keywords — 3 to 7 keywords;
  • seo_type — article;
  • seo_locale — for example, ru_RU, en_US;
  • jsonld — Schema.org markup of the TechArticle type.

The agent follows these rules:

  • it does not change the article title or text;
  • it does not overwrite fields that are already filled in — an employee may have filled them;
  • it does not use the internal part of the article;
  • it does not fill in the html_head field;
  • it relies only on the article content and does not add facts of its own.

All SEO fields are described in Configuring article SEO fields.

Requirements

  1. Swarmica v6.1.0 or higher.
  2. The __ai_mcp_agents feature flag is enabled — see Feature flags FEATURES.
  3. An AI provider that supports AI agents is connected: YandexGPT or ChatGPT. See the current list in The capabilities of various LLM AI providers in Swarmica.
  4. The MCP service is running — see Connecting Swarmica functions to LLMs via MCP.
  5. Administrator rights.

Creating an AI agent

  1. Go to Settings → AI functions → AI agents and click Create.
  2. Select the SEO/GEO/AEO optimization competency.
  3. Enter the agent name — it will be displayed in the interface.
  4. Save the agent.

The agent user and the required access are created automatically.

Automatic runs

Once a day, Swarmica automatically runs the agent for published articles that have at least one empty field among seo_title, seo_description, seo_url, seo_keywords, seo_type, seo_locale and jsonld. Once the fields are filled in, the article no longer gets into the agent queue.

Every agent run consumes AI provider tokens. If the knowledge base has many articles without SEO fields, the initial processing will take several days.

Running the agent manually for a specific article

  1. Go to Settings → AI functions → AI agents and open the agent.
  2. Enter a task, for example: Optimize article 1234.
  3. Click Run.

The progress and status of the run are displayed on the agent page in real time.

Checking the result

  • Open the SEO tab of the article and make sure the fields are filled in correctly. Any field can be edited manually — the agent will not overwrite it.
  • Check the JSON-LD markup with the validation tools — see Configuring Schema.org markup in articles.
  • During content quality checks (AQI), it is recommended to pay attention to whether the SEO fields are filled in and correct.

Related articles

  • Configuring article SEO fields (Open Graph, JSON-LD)
  • Configuring Schema.org markup in articles (JSON-LD, microdata)
  • sitemap.xml and robots.txt for the public knowledge base
  • Enabling indexing of the public knowledge base portal for LLM (llms.txt)