Description

Responsible AI Project, provides a governed and centralized mechanism for developers to quickly build, deploy, and manage generative AI applications for business use cases in a responsible manner.

The Management servcies consists of Model Management, Policy Management and Prompt Management.

• Model Management APIs provides centralized access to Large Language model management.

• Policy Management APIs provides information about client's access to Large Language models. It is coupled with model management capabilitities, guardrails for content control, identity management, and role-based access control.

• Prompt Catalog Management API provides centralized access to prompt template management, grounding rules and system prompts.

Client Benefits

• Internal users at State Street can only utilize a common registry for large language model (LLM), Policy Managements and Guardrail Managements.
• This ensures that all model access is regulated and controlled.

Data Availability

• Data is compartmentalized by application and environment.
• Only client level configurations are stored in RAI.

Description

The retrieval service for RAG (Retrieval Augmented Generation) allows users to choose from different retrievers, providing flexibility in techniques based on usage and requirements.

Client Benefits

The benefits of RaaS include:

• Scalability

• Allowing for plug-and-play integration of different strategies

• Efficiency

• Streamlining the process of information extraction and storage for faster data retrieval

• Flexibility, adaptable to various data types and formats.

Data Availability

RaaS ensures that the parsed and chunked data is readily available for future use, maintaining a structured storage system that facilitates quick access and retrieval of information as needed.

Description

Rag (Retrieval Augmented Generation) as a Service provides the user with the ability to create a knowledge base and manage data sources to efficiently handle their data. It also offers the necessary strategies to leverage RAG (Retrieval-Augmented Generation) as a service.

Client Benefits

Benefits include:

• Efficient data management

• Enhanced decision-making

• Scalability

• Time-saving automation and improved accuracy

Data Availability

Rag as a Service ensures high availability of your data, allowing for seamless access and retrieval at any time. It supports a wide range of data sources and formats, ensuring compatibility and flexibility in managing diverse datasets.

Description

Rag as a Service (RaaS) provides the capability to extract information from data by following a systematic flow of document upload, parsing, and chunking, breaking down documents into smaller pieces for efficient storage and future use. It employs various parsing and chunking strategies that can be easily integrated and customized as per user requirements.

Client Benefits

The benefits of RaaS include:

• Scalability

• Allowing for plug-and-play integration of different strategies

• Efficiency

• Streamlining the process of information extraction and storage for faster data retrieval

• Flexibility, adaptable to various data types and formats.

Data Availability

RaaS ensures that the parsed and chunked data is readily available for future use, maintaining a structured storage system that facilitates quick access and retrieval of information as needed.

Description

ESign Callback API is a RESTful service in State Street to register in vendor (OneSpan) portal to post the events on a transaction like created, expired, signed etc.. ESign callback is representative of live API service in the catalog in terms of data structures, documentation, and behavior.

Client Benefits

• Facsimile data is safe for use in testing processes and automated test cases
• Suitable interface for development of training and documentation

Data Availability

• Data is compartmentalized by environment
• Updates in near real-time
• User-created records are retained.

Description

• The Responsible AI (RAI) Observability Gateway Interface allows users to push LLM conversations used for metrics calculations hosted on AWS.
• The GET API '/trace' can be used with traceId to retrieve the contents.

Client Benefits

• These APIs ensure that the clients are complaint with the RAI Governance Requirements.
• These APIs will help the clients take advantage of the LLM Evaluation Services.

Data Availability

• Data is compartmentalized by application and environment.
• Updates in near real-time

Description

Responsible AI Project, provides a governed and centralized mechanism for developers to quickly build, deploy, and manage generative AI applications for business use cases in a responsible manner.

The Management servcies consists of Model Management, Policy Management and Prompt Management.

• Model Management APIs provides centralized access to Large Language model management.

• Policy Management APIs provides information about client's access to Large Language models. It is coupled with model management capabilitities, guardrails for content control, identity management, and role-based access control.

• Prompt Catalog Management API provides centralized access to prompt template management, grounding rules and system prompts.

Client Benefits

• Internal users at State Street can only utilize a common registry for large language model (LLM), Policy Managements and Guardrail Managements.
• This ensures that all model access is regulated and controlled.

Data Availability

• Data is compartmentalized by application and environment.
• Only client level configurations are stored in RAI.

Description

• The Responsible AI (RAI) Gateway Interfaces allow users to use Chat, Completion, and Embedding APIs of large language models hosted on AWS. These Gateway APIs not only provide access but also ensure the necessary authorizations and policy checks before making calls to the large language models.

Client Benefits

• Internal users at State Street can only utilize a common registry for large language model (LLM).
• This ensures that all model access is regulated and controlled.

Data Availability

• Data is compartmentalized by application and environment.
• GenAI calls and responses are made in real-time.

Description

Allows the client to view position data searchable by fund, client ID and date. Position quantities are broken down into granular balance buckets including shares available, blocked, borrowed, collateral in/out, on loan, pledged, restricted and more.

Client Benefits

• Transparency into your custody security balances assisting in the validation availability to support trade settlements.
• Increased efficiency and reduced query volumes through reduction of manual and email-based operations.
• Tailored API requests which allow for customized data extracts.

Data Availability

• The API will provide date driven custody positions as it gets updated from State Street Custody applications.
• The API will provide the custody positions from the core authoritative source as available on State Street Custody applications.
• The API will provide the custody position data for entity IDs in which the consumer is entitled.

Description

Investment Risk Analytics API provides a range of analytics metrics and aggregations to be used for market risk analysis and regulatory reporting purposes of State Street clients.

Client Benefits

• A new convenient way to provide market risk data to State Street clients through API.
• Increased efficiency and reduced query volumes through reduction of manual and email-based operations
• Flexible method to query market risk metrics in an ad-hoc manner.
• Different filtering methods with a use of available input parameters.

Data Availability

• Data is updated as per the frameworks' requirements - daily/monthly/quarterly.
• Data is structured in endpoints to address various market risk frameworks' requirements.