Semantic article search for a specialist media title

A SaaS B2B business working with Cloud Combinator on AWS. The client is anonymised at their request.

The proof of concept had four focus areas, each shaping how readers would experience the new search.

The client articles are ingested from WordPress into an Amazon S3 repository, split into article content and metadata such as category, author and publish date, with automatic syncing so new and updated content flows in. That bucket is the primary data source for an AWS Q Business application, which indexes the corpus for semantic retrieval within the London region.

The proof of concept demonstrates that AWS Q Business can deliver intent-aware, premium-gated article discovery over the client's archive, with acceptance measured against the project success criteria and the AWS environment shown to be suitable for continued development. The performance and relevance targets below are proof-of-concept success criteria, validated through sample testing.

• Intent-aware discovery — AWS Q Business semantic search returns three or more genuinely related WordPress articles per query, with relevance validated by sample testing.
• Tailored to how people read — Three persona prompts, skim reader, in-depth researcher and full-article consumer, so responses match reader intent.
• Premium gating — IAM Identity Center federation and role-based access restrict the feature to premium subscribers, with no leakage to free users.
• Compliant by design — GDPR-aligned data handling with encrypted, access-logged S3 storage for the indexed archive and CloudTrail auditing of data flows.
• Handover — CloudFormation templates, system prompt files, IAM role definitions and architecture diagrams, closed out with a live demo.

With the proof of concept proving out the architecture, the client has a validated AWS foundation to move semantic search into production and build further reader-facing intelligence on top of its archive.