Orca turns natural language into cloud security queries with Bedrock
Discovery AI converts plain-language questions into Sonar filters. AWS says the system cut query cost by 90%, input tokens by 93% and latency by 75% against its initial design.
Source review period: 2026-10-13
Published: · Updated:
The source does not state the event date; the period shown is the source review period.
Orca Security worked with AWS on Discovery AI, a natural-language interface to its Sonar security graph, which covers more than 900 cloud and application resource models. The tool passes a user’s question through a schema-aware pipeline to produce valid Sonar queries, with steps for choosing relevant schemas, retrieving examples and validating the resulting filters.
AWS’s October 13 technical post says the first prototype placed the entire schema in one large prompt, using about two million input tokens per workload and taking roughly 60 seconds to respond. The production system handles thousands of queries a day at about 15 seconds per response. AWS reports reductions of 93% in token use, 90% in cost and 75% in latency compared with that initial version; these are Orca’s reported internal comparisons.
Data
AWS technical post published
AWS described Orca Discovery AI’s conversion of natural-language security questions into Sonar filters on October 13.
- · Occurred
AWS technical post published
AWS described Orca Discovery AI’s conversion of natural-language security questions into Sonar filters on October 13.
Methodology
The date is the publication date shown by the AWS primary source; no launch or deployment date beyond the source is inferred.
Updated:
Subjects and places
Subject proposals
- Artificial intelligenceSuggested
- Subject suggested from the title or summary: AIaws.amazon.com ↗