AI News

Professional services firm Booz Allen Hamilton recently introduced aiSSEMBLE Baseline, an open-source iteration of its aiSSEMBLE solution, which aims to streamline the process of developing, engineering and deploying AI solutions.

The company’s holistic, lean-based approach for designing, developing and fielding AI solutions spans the engineering lifecycle – from data processing (DataOps), to building, training and tuning models(MLOps), through secure operational deployment (DevSecOps).

The aiSSEMBLE Baseline is Booz Allen’s foundational software integration toolkit that implements and automates the company’s government tailored MLOps approach as a series of reusable, flexible and extensible software components.

“This enables rapid development of fit-for-purpose solutions meeting unique mission needs via a common software baseline that maximizes quality, ensures consistency, and reduces delivery risk,” said Bryan Castle, director of software engineering for aiSSEMBLE at Booz Allen.

Using aiSSEMBLE Baseline, organizations can streamline the delivery of AI at an enterprise level, addressing the challenge of a “pilot purgatory” and helping organizations avoid vendor lock-in.

The solution offers foundational code and open-source documentation, available for free use by government, academic, non-profit and commercial entities.

Organizations can utilize aiSSEMBLE via existing internet mechanisms, eliminating the need for VPNs.

Castle explains the solution employs reusable software components to standardize AI solution development.

“The open architecture allows clients to choose the optimal tools and components based on their specific mission requirements,” Castle explained.

It includes default configurations but can be readily adapted to support a variety of open source software (OSS), government off the shelf (GOTS) and commercial off the shelf (COTS) software components.

He said the goal is to provide clients with real flexibility versus locking them into a proprietary technology stack and recurring cost model and added that as an open source solution, aiSSEMBLE can provide federal agencies with two primary benefits.

“First, it codifies best practices for AI development to accelerate adoption and scale implementations,” he said. “Second, it supports an open architecture, so it is extensible to support future innovations.”

Castle said the strategy behind releasing it as an open source solution would help accelerate federal AI adoption.

“It also provides our clients with assurances about adopting the technology while enabling us to continue to develop our own innovative, differentiated solutions with the technology,” he said.

The platform is also designed to facilitate collaboration by allowing vendors to integrate with Booz Allen’s AI delivery approach.

“Through our support across the U.S. federal government, we have established numerous best practices which are codified into the aiSSEMBLE software baseline,” Castle said. “This means that new projects will take advantage of these best practices to accelerate solutions and increase the probability of success.”

Booz Allen plans to leverage aiSSEMBLE in various projects and supports a portfolio of projects across its civil, defense and national security sectors, including support for a recent award from the National Oceanic and Atmospheric Administration (NOAA).

In this initiative, aiSSEMBLE will provide the framework for data processing and dissemination of weather data, enhancing operational workflows.

“Each project is unique in their mission needs, but aiSSEMBLE provides a foundation of common components that are accelerators for preparing data, building AI models and deploying enterprise AI solutions,” Castle explained.

The release comes as private and public organizations are struggling to make progress and devise implementation of AI, among the myriad other challenges they face as waves of digitalization transform the business landscape.

Technology giant Salesforce has also doubled down on a host of AI-aided tools, expanding its Mulesoft Intelligent Document Processing (IDP) module with AI capabilities to simplify data extraction and organization from various formats.

This integration includes the deployment of Einstein AI across multiple platforms, such as Mulesoft IDP, Flow Builder, and Anypoint Code Builder, enabling prompts for invoking large language models (LLMs) via natural language interfaces for generative and predictive AI functions.

The company also enhanced its integrated development environment (IDE) by introducing a configuration panel and providing access to accelerators, templates in Anypoint Exchange, and integration with external version control systems (VCS) to facilitate easier management of updates.