Lucius AI Slashed Tender Search Latency to 24 Milliseconds
The procurement platform improved search speeds by moving its vector and relational data to a consolidated database architecture.
Updated on Sept. 30, 2026 in Artificial Intelligence
Live Poll
Do you trust automated AI agents to handle critical data management for your business?
Lucius AI has migrated its public procurement search platform to Google AlloyDB for PostgreSQL, resulting in a reduction of semantic search query latency from 1.14 seconds to 24 milliseconds. The platform currently tracks over 210,000 tenders across international markets using Google Gemini models.
Why it matters
By consolidating relational, vector, and audit log data into a single system, Lucius AI removed the need for separate infrastructure products. This shift enables the platform to support complex procurement workflows with fewer dedicated data engineering resources.
The system achieved a mean latency of 77 milliseconds for retrieval reranking and processed 115,820 records for its semantic index in 10.6 minutes. The migration utilized a ScaNN index for vector comparisons alongside a Model Context Protocol to link AI agents directly to the database.
The players
Lucius AI
A procurement technology firm that utilizes generative AI to analyze international tender data and provide bid recommendations.
A major cloud provider that offers the AlloyDB for PostgreSQL database and the Gemini large language model suite.
The details
The platform functions by integrating relational databases with vector embeddings—numerical representations of data that allow for semantic similarity searches—within a unified environment. An AI agent communicates with the database using the Model Context Protocol, a standard for connecting AI models to data sources, allowing for the generation of compliance matrices and bid recommendations. This architecture replaces the previous setup of siloed products, facilitating efficient data access across European and Australian production regions.
Timeline
September 30, 2026: Article publication date.
The Tech Race
The adoption of integrated vector database features in platforms like AlloyDB represents a competitive effort to minimize data infrastructure complexity for AI-driven applications. This development places Lucius AI among a growing set of firms replacing distinct vector databases with consolidated, high-performance PostgreSQL-compatible backends.
Users of the Lucius AI platform will experience faster search results and more responsive bid recommendation generation during the procurement process. These performance gains apply to the full catalogue of 210,000 tenders sourced from 13 different regions globally.
The takeaway
The successful migration demonstrates that consolidating vector embeddings within a traditional relational database can significantly lower latency for complex AI queries. Watch for further platform updates as Lucius AI potentially integrates additional data sources into this unified architecture.
Further reading
For more on the infrastructure supporting large-scale enterprise models, explore our Artificial Intelligence section.
Source note: This article includes information reported by IT Brief Australia.
Live Poll
Do you trust automated AI agents to handle critical data management for your business?
