Soft-land readers who hit a missing URL by recommending the post they were probably looking for, based on the URL slug itself.
How it works
Smart 404 turns the WordPress 404 page from a dead end into a recovery surface. When a request comes in for a URL that no longer exists, ClassifAI converts the missing slug to an embedding, runs a vector search against the embeddings of every published post on the site, and displays the closest matches as did you mean… recommendations. Because the lookup is semantic, even a heavily mangled URL — or one that points to a post structure that has been retired — usually resolves to something useful. The feature is delivered through a block you place in the 404 template.
Configuration
- Number of suggestions returned per 404.
- Post types in scope for the vector search.
- Provider and model selection.
- Allowed roles and an allowed-users list for granular access control.
Providers
- OpenAI Embeddings
- Azure OpenAI Embeddings
- Ollama Embeddings (locally hosted)
Requirements
Vector search requires ElasticPress to be installed and active — ClassifAI uses ElasticPress’s vector index as the search backend. Teams running an entirely on-prem stack can pair ElasticPress with the Ollama embeddings provider so 404 lookups never leave their network.
Use cases
- Sites mid-way through a content reorganisation, where slugs have changed faster than redirects can be authored.
- Long-running publications whose archive URLs have been mangled by years of platform migrations.
- Marketing pages frequently shared on social with truncated or typo’d URLs.