Product descriptions:
Hawaiian Electric Company delivers energy to over 95% of Hawaii’s population, operating in a complex regulatory environment that requires precision, speed, and trust. For decades, legal and regulatory teams have spent hours searching for critical details hidden across more than 60,000 filings, rate cases, and technical appendices, creating bottlenecks and inconsistent responses. By deploying Databricks Agent Bricks and Knowledge Assistant, Hawaiian Electric unified its entire regulatory corpus, automating search and response for filings, inquiries, and legal requests. The trusted AI assistant now delivers verifiable answers in minutes, slashing hours of manual research and setting a new bar for regulated utilities.
Unlocking Business Value from Decades of Unsearchable Documents
Every rate change, compliance inquiry, and regulatory submission at Hawaiian Electric depends on the fast and precise referencing of historical filings and precedents. But with documents scattered across legacy systems and storage, staff often spent hours searching, and teams rarely spoke with a consistent voice. “It could take hours just to locate a single reference,” explained Joel Wasson, Manager of Enterprise Data Analytics. Different team members sometimes interpreted information differently, which slowed response quality and regulatory consistency.
Earlier attempts included a standalone desktop search tool and a homegrown solution stitched together from multiple components — but these still demanded heavy manual development, offered limited governance, and couldn’t scale as new data sources emerged. “We needed a smarter, more unified approach — a system that could enforce governance and extract answers with context and citation,” said Emmanuel “Manny” Alboroto.
Trusted AI, Built on Secure, Governed Data
Using Databricks Agent Bricks, the team built a Knowledge Assistant tailored to its 60,000-document regulatory library. Now, legal and regulatory teams retrieve filings, historical opinions, and precedent cases at three times the previous speed — complete with in-line citations and direct download links for compliance and verification. “The ability to immediately validate answers isn’t just a feature — it’s what drives real trust,” Manny stated.
Governance is foundational: by leveraging Unity Catalog, Hawaiian Electric enforces access controls and keeps all data, models, and permissions inside its cloud boundary. “Data governance is non-negotiable for us,” said Alan Hong, Senior Data Systems Analyst. “With Databricks, we have a single, secure environment connecting data lineage and user permissions end to end.”
Quality control is an ongoing process: Hawaiian Electric combines human-in-the-loop review with Agent Bricks Knowledge Assistant and evaluation metrics. The team’s first prototype improved response accuracy from 30% to 85% — a 3X leap that improved trust with both legal groups and external regulators. Early testing reveals growing user confidence and an expanding library of validated answers aligned with evolving legal standards.
Hours Saved, Consistency Achieved, and Confidence Gained
What previously took hours can now be accomplished in minutes. Legal and regulatory teams use natural-language questions to instantly find grounded, cited answers, reducing multi-hour searches to under five minutes. This shift enables staff to focus on substantive, strategic work rather than low-value document retrieval. “It’s helped us establish a common voice across teams,” Joel reported, “so everyone starts with the same factual baseline instead of re-interpreting documents independently.”
The AI assistant has also driven a culture of trust and transparency. Staff rely on in-tool citations and document links to validate outputs, transforming a former “black box” model into a transparent, collaborative engine. “That ability to double-check the source is what gets everyone on board,” Manny noted from the regulatory team’s response.
Looking forward, Hawaiian Electric is expanding agent-driven solutions into new domains, including a cloud-based geospatial data lakehouse to support maintenance, resilience, and digital twin initiatives. With tools like Knowledge Assistant, AI/BI Genie, and Supervisor, teams envision a future where anyone — technical or not — can interact with real-time data through natural language, driving value and innovation across the utility.