Digital Regulatory Intelligent Filing Tool (DRIFT)
Digital Regulatory Intelligent Filing Tool (DRIFT)
Smart Filing for Smarter Compliance
DRIFT GenAI Platform for Digital Regulatory Automation helps utilities streamline the time-intensive process of preparing regulatory filings. By leveraging generative AI and AWS accessibility, DRIFT can generate filing content and structure, suggest supporting data arguments and interrogatories, and continuously scan public and approved regulatory filings to surface the most relevant and up-to-date regulatory support.
Eliminates need for expensive model fine-tuning while contextualizing requests with necessary data sources.
Source documents are cited down to the specific page, minimizing hallucination risks.
Solution’s outputs have been optimized for utility-specific use cases, with proprietary translation input from utility wildfire experts.
Complex prompt engineering is abstracted to improve accessibility of the tool for end users.
Why DRIFT?
Engineered File Ingestion
Add to collection as more data becomes available.
Scalable: nearly unlimited document capacity, vs. ChatGPT limit of 20 documents.
High-quality OCR (Optical Character Recognition) with custom pre-processing of indexes.
Why DRIFT?
Private, High Quality Data Collections
All data processed by the tool remains within the owner’s AWS account and access is controlled by AWS IAM permissions.
Curated data sets allow model to provide detailed responses past what is available in LLM base knowledge.
Collections enable isolated sets of documents to serve distinct purposes.
Why is DRIFT secure and safe GenAI?
There are three main reasons why DRIFT is a safe tool for utilities to engage with:
Badget Text
Document Search & Citation
Reference documents are cited as part of each system response, ranked by relevance.
The individual pages leveraged by the model are displayed along with their metadata.
Users can filter their responses to only leverage documents matching specific criteria, such as state and year.
Badget Text
Code Base & Architecture
Source code and document collections are stored within the AWS environment in which DRIFT is deployed.
Content uploaded into DRIFT remains within the AWS account.
DRIFT enables an audit trail of prompts run and files uploaded by user.
Badget Text
Public vs. Private Document Curation
Users of DRIFT determine what content is leveraged by the tool, rather than relying on data available on the public internet.
Supports published and private collections:
– Private: documents can only be managed and accessed by the individual user that created them
– Published: documents available to all users within DRIFT