
BidiTech has officially launched the AquaBrain™ Water AI Model Platform. Designed for water utilities, treatment plants and pumping stations and deployed on premises, the platform integrates a large language model, a domain knowledge base and operating data analysis in a single platform, with all data and models running on the customer's own servers.
Why an On-Premises Water AI Model
Water operating data and internal documents are highly sensitive. General-purpose cloud LLMs lack water-sector expertise and struggle to meet data compliance requirements. AquaBrain™ deploys the LLM, vector search and knowledge base inside the utility, with separate workspaces for the group and each plant, so data stays within the domain and access is controlled.
Four Core Capabilities
- Water knowledge base: operating procedures, standards, equipment documentation and technical books are ingested in one place; formulas and tables are parsed as in the original, and search results are traceable to their sources.
- Knowledge Q&A: answers are based primarily on the knowledge base with sources cited; where no basis exists, the system says so explicitly and does not fabricate standard clauses or limits.
- Operational diagnostics: diagnoses are structured as symptom confirmation, possible causes, troubleshooting steps, recommended actions, and safety and reporting.
- Operating data analysis: the analysis engine precisely calculates energy consumption, specific energy consumption and pump combination efficiency, while the LLM interprets the results and drafts operations analysis reports.
Application Example
At a lift pumping station, AquaBrain™ analyzed approximately 26,000 minute-level operating records from 5 pumps over 18 days. The most energy-efficient pump combination had about 13% lower specific energy consumption than the usual duty combination but accounted for less than 4% of actual operating time; two fixed-speed pumps running at mains frequency had a power factor of only 0.71–0.76; and a frequency signal conversion error was found for one pump. These findings translate directly into scheduling optimization, reactive power compensation and data governance measures.
From Knowledge to Control
AquaBrain™ shares its design foundation with BidiTech's MPC technology and the DeepControl™ Edge Controller. Control models for pump scheduling, dosing and aeration validated on the platform can be deployed to edge controllers at plants and stations.
Water utilities and partners are welcome to book a demonstration: provide a sample of operating data and experience the complete analysis and Q&A workflow on site.