Oracle has introduced a new search tool called Trusted Answer Search. It aims to deliver accurate and trustworthy results for large organizations. Instead of using traditional language models, it relies on vector search to find relevant documents within a controlled set of approved files.

How Trusted Answer Search Works

The system works by having companies define a curated “search space” of approved reports, documents, or application endpoints. These are paired with metadata to help organize the content. When a user inputs a natural language query, the system uses vector-based similarity matching to find the most relevant pre-approved document.

Unlike systems that generate responses from raw text, Trusted Answer Search deterministically maps the user’s question to a specific “match document.” It then extracts any needed parameters and returns a structured, verifiable outcome, such as a report, URL, or an action. This approach makes the results more predictable and easier to audit.

Benefits and Feedback Loop

This system also includes a feedback feature, allowing users to flag incorrect matches and specify what the correct result should be. This feedback helps improve accuracy over time. Oracle’s Tirthankar Lahiri explained that many enterprises need more deterministic natural language query systems to avoid inconsistent answers and meet compliance standards.

Independent consultant David Linthicum noted that companies valuing predictability over creativity could find this tool especially useful. He pointed out that industries like finance and healthcare, which are heavily regulated, could benefit from the transparency and auditability Trusted Answer Search offers.

Trade-offs and Challenges

While the technology reduces costs associated with running large language models, it shifts spending toward data management. Organizations need to invest in curating documents, designing taxonomies, obtaining approvals, and ongoing maintenance. Linthicum highlighted that document curation and continuous tuning are essential for optimal performance.

However, keeping curated data up-to-date presents challenges. Scott Bickley from Info-Tech Research Group warned that as the volume of external data grows—such as regulatory updates or supplier certifications—maintaining accurate and current results becomes harder. There’s a risk of delivering plausible but incorrect answers, especially when documents contradict each other or use similar language in different contexts.

Despite these concerns, Lahiri from Oracle believes that Trusted Answer Search can mitigate some risks. Because it retrieves content from trusted, parameterized documents rather than relying solely on static data, it can adapt more flexibly to changing information. This makes it a promising option for enterprises seeking reliable, compliant search solutions.

Inspired by

• https://www.infoworld.com/article/4160436/oracle-delivers-semantic-search-without-llms.html

Sources

• infotech.com
• hfsresearch.com
• oracle.com