How to Automatically Tag and Categorise WordPress Posts With AI
You can automatically tag WordPress posts with AI using Krom Automation, a free WordPress plugin. The AI Auto-Tag action is included in the free version, no Pro licence required.
You connect a “Post Published” trigger to the AI Auto-Tag action, supply your own API key from OpenAI, Google Gemini or Groq, and every new post gets tagged the moment it goes live. The whole setup takes under 10 minutes.
The free plugin handles tagging for new posts on publish. If you need to retroactively tag hundreds or thousands of existing posts, the most practical approach combines a scheduled workflow with a Post Updated trigger, which lets you process your back catalogue in controlled batches. We cover both flows below, along with the tag governance decisions that every other guide skips entirely.
This guide also addresses the question nobody answers directly: how many tags should AI assign, what happens to your existing tags, and how do you stop a well-meaning automation from turning your tag archive into 400 near-duplicate terms that confuse both readers and search engines.
Browse the full feature list to see every free trigger and action before you commit to a setup.
What the AI Auto-Tag Action Actually Does
The AI Auto-Tag action reads the post content and uses your chosen AI model to generate relevant tags. It then applies those tags to the post inside WordPress, the same way you would add them manually from the editor. No markup is altered, no post content is modified, and the action leaves existing tags in place by default.
The critical behaviour to understand before you build a workflow: by default, the action adds tags rather than replaces them. If a post already has three manually set tags, the AI adds its suggestions alongside them.
This is the right default for most sites. If you want the AI to own tagging entirely and ignore any manual tags, that is a deliberate choice you make in the action settings, not an accident.
Tag creation behaviour also matters. The action can be configured to:
- Use existing tags only, so the AI matches content to tags already in your database and never creates a new term
- Create new tags freely, so the AI generates whatever terms fit the content, even if those terms do not yet exist in WordPress
- Cap the maximum tag count per post, which we recommend setting to between 3 and 7 to prevent tag sprawl
For most editorial sites, “existing tags only” is the safer starting point. You define the vocabulary once, the AI applies it consistently, and your tag archive stays clean. Switching to “create new tags” makes sense when you are building a new site and want the AI to help you discover what your tag taxonomy should be, not when you are managing an established archive.
Letting AI create tags freely on an established site is like hiring a filing clerk who invents their own categories. The filing gets done. Finding anything afterward takes twice as long.
Which AI Providers Work and What Do They Cost You
Krom Automation never charges per AI call and never marks up tokens. You bring your own API key, pay your provider at their standard rate, and the cost goes directly to them. wpRigel takes nothing from that transaction.
The free plugin supports seven models across three providers:
- OpenAI: GPT-4o Mini, GPT-4o
- Google Gemini: Gemini 2.0 Flash, Gemini 2.5 Flash Preview, Gemini 2.5 Pro Preview
- Groq: Llama 3.3 70B, Llama 3.1 8B
For auto-tagging specifically, GPT-4o Mini and Gemini 2.0 Flash are the cost-effective choices. A typical 1,000-word post generates a tagging request of roughly 1,000 to 1,500 tokens in and 20 to 50 tokens out.
At GPT-4o Mini rates, tagging 500 posts costs well under $1. If you are processing a large back catalogue, Groq’s Llama models are free on the free tier up to their rate limits, which makes retroactive bulk tagging on a budget genuinely achievable.
Setting Up Auto-Tagging for New Posts: Step by Step
Before you build the workflow, read how Krom Automation works if this is your first workflow. The core model is simple: a trigger fires when something happens, actions run in response, and the canvas connects them visually.
To tag every new post on publish:
- Install and activate the free Krom Automation plugin from the WordPress.org plugin directory
- Go to Krom Automation in your WordPress admin and click “Add New Workflow”
- On the canvas, add the Post Published trigger and set the post type to “Post” (or whichever post type you want to tag)
- Add the AI Auto-Tag action and connect it to the trigger node
- In the action settings, choose your AI provider, paste your API key, and set your maximum tag count and tag creation mode
- Save and activate the workflow
The next post you publish will be tagged automatically within seconds of the trigger firing. Use the workflow simulator to run a dry test against an existing post before you go live. It shows exactly what tags would be applied, with zero side effects on real content.
If you want to add a human review step, you can insert a conditional branch after the AI Auto-Tag action. See the conditions and branching documentation for how to build Yes and No paths, which lets you route borderline posts to a draft status for manual review while publishing clean ones immediately.
How to Bulk Auto-Tag Existing Posts
Every other auto-tagging guide ignores your existing content. Most sites have hundreds or thousands of posts that predate any AI tool, and those posts are the ones with the most search traffic at stake. Here is a practical approach for processing them without breaking your site or your tag taxonomy.
The core problem with bulk retroactive tagging is volume control. Sending 2,000 posts to an AI provider at once is not a workflow, it is an invoice. The safer approach is batching: process 20 to 50 posts per run, on a schedule, until the back catalogue is done.
In Krom Automation Pro, the Schedule Trigger lets you fire a workflow hourly, daily or on a custom interval. Combined with a query that fetches untagged posts and passes them through the AI Auto-Tag action, you get controlled batch processing with full execution logs and failure notifications if anything goes wrong.
For the free plugin, the practical equivalent is to export a CSV of untagged post IDs, bulk-edit them in batches inside WordPress to trigger a “Post Updated” event, and let the workflow catch each update. It is more manual, but it works and costs nothing beyond the API calls.
Retroactive tagging is not a one-afternoon job. It is an editorial project. The sites that do it well set a tag limit, audit the results after the first 100 posts, and adjust the prompt before running the rest.
Tag Governance: The Part Every Other Guide Skips
Tag sprawl is when your tag archive slowly fills with near-duplicate terms: “marketing strategy”, “marketing strategies”, “content marketing strategy”, “digital marketing tips”. Each term has one or two posts attached.
None of them rank for anything. All of them generate thin archive pages that dilute your site authority.
AI auto-tagging accelerates whatever tagging discipline you already have. Good discipline scales well. No discipline scales into a mess faster than you can clean it up.
Before you run any auto-tagging workflow, answer these three questions:
- What is the maximum number of tags per post? We recommend 3 to 7. More than 7 and the tags stop being meaningful signals. Fewer than 3 and you are not getting the taxonomy coverage the feature exists to provide.
- Who owns the tag vocabulary? If the AI creates new tags freely, nominate someone to audit new terms weekly. A monthly audit is too infrequent once you have a high-volume site.
- What is your merge strategy for near-duplicates? WordPress does not merge tags automatically. You will need a taxonomy management plugin or manual cleanup when you spot drift. Building the cleanup step into your governance workflow from day one is far cheaper than a one-off cleanup of 600 tags six months from now.
The “existing tags only” mode in the AI Auto-Tag action is the best technical safeguard. If the AI can only choose from the 80 tags you have already defined, the worst it can do is pick a slightly imprecise match.
That is recoverable. A freely-generating AI choosing new tag names daily is not.
Will AI Auto-Tagging Help or Hurt Your SEO?
The honest answer is: it depends entirely on how you configure it, and the risk comes from the output rather than the mechanism.
Tags in WordPress generate archive pages. Each archive page is a URL that Google crawls and indexes.
A tag with 30 posts behind it is a meaningful topical signal. A tag with one post behind it is a thin page with near-zero ranking potential, and if you have 200 of those, you are giving Google 200 reasons to view your site as shallow.
The configuration choices that protect your SEO:
- Set the tag creation mode to “existing tags only” until your taxonomy is mature
- Cap tags per post at 5 or fewer during the initial rollout
- Use the execution logs in Krom Automation to review the first 50 tagged posts before running the full back catalogue
- Add noindex to tag archives with fewer than 5 posts, using Yoast SEO or RankMath’s taxonomy settings, until the archive builds enough content to justify indexing
Done this way, AI auto-tagging is a genuine SEO lever. Consistent tagging across hundreds of posts creates topical clusters that manual tagging rarely achieves because it is too easy to forget a tag or use a slightly different term on each post.
Does This Work With WooCommerce Products and Custom Post Types?
Yes. In the Post Published trigger, you set the post type. WordPress registers WooCommerce products as the “product” post type.
Set the trigger to “product” and the AI Auto-Tag action will tag products exactly as it tags posts. The same applies to any custom post type registered on your site.
If you run WooCommerce, you may also find the WooCommerce Subscriptions integration useful for broader order and subscription automation beyond tagging. And if your site uses forms to gather data that informs content creation, the Gravity Forms integration and WPForms integration let you trigger other workflows from form submissions in the same canvas.
The Timing Limitation You Should Know About
Krom Automation runs background tasks via Action Scheduler, the same queue system WooCommerce uses. This means the AI Auto-Tag action does not run during the page load that saves your post. It fires in the background, typically within a few seconds on a site with normal traffic.
On very low-traffic sites, WP-Cron only fires when someone visits the site. If your site goes several hours without a visitor, a scheduled or queued action waits until the next visit before it runs.
The fix is to disable WP-Cron in wp-config.php and configure a real server cron job to call the WordPress cron endpoint every minute. This is a one-time server configuration that costs nothing and makes every time-dependent automation on your site more reliable, not just Krom Automation.
For a broader view of how AI can power WordPress automation beyond tagging, see our guide on using AI to automate your WordPress site.
Free vs. Pro: What Do You Actually Need for Auto-Tagging?
| Use case | Free plugin | Pro required |
|---|---|---|
| Tag new posts on publish | Yes | No |
| Tag WooCommerce products on publish | Yes | No |
| Tag custom post types on publish | Yes | No |
| Conditional branch after tagging (review step) | Yes | No |
| Dry-run testing with the simulator | Yes | No |
| Scheduled batch processing of existing posts | No | Yes (Schedule Trigger) |
| AI workflow generation from plain English | No | Yes |
| Incoming webhook to trigger tagging from external events | No | Yes |
The free plugin handles the primary use case fully. Pro becomes relevant when you want to run scheduled batch jobs against existing content or trigger tagging from external systems via webhook.
What It Costs Across One, Five and Twenty Sites
| Scenario | Free plugin only | Krom Pro Basic (1 site) | Krom Pro Standard (5 sites) | Krom Pro Enterprise (unlimited) |
|---|---|---|---|---|
| 1 site, new posts only | $0/year | From $69/year | From $149/year | From $299/year |
| 5 sites, new posts only | $0/year | 5x Basic licences | From $149/year | From $299/year |
| 1 site, includes batch back-catalogue | Not available | From $69/year | From $149/year | From $299/year |
| 20+ client sites, agency use | $0/year, manual setup each | Not practical | Not practical | From $299/year |
| Any tier, lifetime option | Free forever | From $199 once | From $399 once | From $699 once |
Launch pricing is currently applied. See the current pricing for up to date figures. All Pro plans include a 14-day money-back guarantee and every Pro feature, with the only difference between plans being the number of site activations.
For context, at the time of writing, Uncanny Automator’s automation plans start at $199 per year for one site, with AI features starting at $300 per year. Krom Automation’s free tier includes AI actions at no additional cost from wpRigel, which is a material difference for teams watching per-feature costs.
For agencies running 10 or more client sites, the Enterprise lifetime plan at $699 one time competes directly with a single year of per-site SaaS tooling. That is the calculation worth running before you renew anything.
How This Compares to Standalone Auto-Tagging Plugins
There are WordPress plugins built specifically for AI tagging. TagPilot AI uses the Dandelion NLP API and focuses on the install-and-forget flow.
Qwebmaster AI Auto Tagger takes a privacy-first approach where all processing happens on-server with no external API. PublishPress Taxonomies Pro integrates with OpenAI and IBM Watson for taxonomy management at a broader level.
The practical difference with Krom Automation is scope. A dedicated tagging plugin does one thing. Krom Automation does tagging as one of 21 free actions inside a workflow that can also send emails, update post meta, create users, make HTTP requests, and moderate comments.
If tagging is the only thing you need, a dedicated plugin is a legitimate option. If you are already building automations or plan to, adding a tagging step to an existing workflow costs nothing extra.
The other difference is the audit trail. Krom Automation logs every execution with a per-step breakdown, so you can see exactly which tags were applied to which post and when. Debugging a misfiring tag rule in a dedicated plugin typically means searching through a settings page.
In Krom Automation, you open the execution history and read the result directly. Our guide on AI comment moderation shows the same pattern applied to a different content management problem.
Also from wpRigel
Pollify is wpRigel’s Gutenberg-native poll, survey and quiz plugin. Polls are built directly inside the block editor as real blocks, so there are no shortcodes to paste and no separate interface to learn. It is the fastest way to add audience engagement to any post or page without leaving the editor.
Commandify is a command palette for the WordPress admin. Press Cmd or Ctrl plus K from anywhere in the admin to jump to any page, search any content type, and run admin actions without clicking through menus. It is the only command palette plugin with genuine WooCommerce depth, covering orders, products and customers as first-class commands.
Our Verdict: Who Should Use This and Who Should Not
Use the AI Auto-Tag workflow if you publish more than 4 posts per month and find tagging inconsistent across your archive. Inconsistency is the real cost: posts on the same topic tagged differently create fragmented archives that neither readers nor search engines can navigate. The free plugin handles this with zero recurring cost from wpRigel, and the only ongoing expense is your AI provider’s token fees, which are negligible at typical publishing volumes.
Do not use it if your site has fewer than 20 posts total. The overhead of setting up a workflow and governing a tag taxonomy is not worth it at that scale.
Manual tagging takes minutes and gives you full control when the site is small enough for one person to manage in their head. Come back when you are publishing regularly and the manual process starts creating gaps.
For agencies managing multiple client sites, the workflow is worth setting up once, exporting as JSON, and importing to each new site. Krom Automation supports workflow import and export, so you build the tagging workflow once and deploy it everywhere. That is the use case where the Enterprise lifetime plan earns its cost back in the first quarter.
Download the free plugin from the WordPress.org plugin directory and set up your first auto-tagging workflow today. If you decide you need scheduled batch processing for existing posts, compare all three Pro plans to find the right fit.
Frequently Asked Questions
Will the AI Auto-Tag action overwrite my existing tags?
No. By default it adds tags alongside any existing ones.
If you want the AI to replace all existing tags, that is a deliberate setting you change in the action configuration. The default protects manually set tags.
How many tags should AI assign per post?
We recommend capping at 3 to 7 per post. Below 3, the taxonomy coverage is too thin to build meaningful topical clusters. Above 7, the tags start to overlap and lose their value as navigational signals for both readers and search engines.
Does auto-tagging work with custom post types, not just standard posts?
Yes. In the Post Published trigger, you select which post type fires the trigger. Any custom post type registered on your site, including WooCommerce products, is available in that dropdown.
Is there a free AI auto-tagger for WordPress, or do I always need to pay for an API?
Krom Automation is free to download and the AI Auto-Tag action is included at no cost from wpRigel. You do need an API key from OpenAI, Google Gemini or Groq. Groq’s Llama models have a free tier that is sufficient for moderate tagging volumes, making it possible to start with no API cost at all.
Can I auto-tag existing posts, or only new ones going forward?
The free plugin tags posts when they are published or updated, so you can trigger retroactive tagging by batch-updating existing posts. Krom Automation Pro adds a Schedule Trigger that lets you run automated batch jobs against existing content on a set interval without manual edits.
Will AI-generated tags help my SEO or create thin archive pages?
Both outcomes are possible and the configuration determines which one you get. Setting the action to “existing tags only” and capping at 5 tags per post creates consistent topical clusters that support SEO. Allowing the AI to create new tags freely without governance creates thin archive pages quickly.
The technical tool is neutral. The editorial discipline around it is what makes the difference.