AI IQ

The scope statement defines the product, service, and results to be delivered by a project. The scope statement outlines the boundaries for the project work and provides the input to create the budget and schedule. AI is applied to scope for two primary purposes: to help create the scope statement and to review the content for consistency, checking for errors and omissions.

As Professor Bent Flyvbjerg points out in his book How Big Things Get Done, in an analysis of 258 projects with a budget over $1 billion, only 8.5 percent met the budget and schedule, and a minuscule 0.5 percent were delivered on time, within the budget and achieved the expected benefit. These results suggest that the project methodology needs to be reconsidered. Fortunately, new technology, specifically AI, offers hope for improving project results, and it starts with the scope statement. Both agile and waterfall projects can benefit.

Agile

For agile projects, a study by Accenture found that the root cause of 35 percent of defects in production was due to errors in the requirements document. Based on the wording of user stories, natural language processing (NLP) tools find errors and omissions. For example, the requirements might define the details of a feature, but that feature is never tested. Alternatively, there may be testing for a feature that is only vaguely defined. AI-based scope review algorithms find these inconsistencies up to 20 times faster than a human and are especially useful for requirements that are hundreds of pages long. This is not a distant reality or exaggeration. A government department recently created a software consolidation project, moving the functionality from a variety of different applications into a single software solution. A previous similar project was significantly over budget and schedule. However, the main issue was that it did not perform as expected, creating negative public criticism. For the current software consolidation project, the government acquired an AI-based NLP tool to thoroughly review and correct the user stories before starting the project.

Waterfall

Construction projects have two areas where AI can help provide a more accurate scope document. First, a draft scope document can be automatically created using a large language model (LLM) or a database of previous documents. This saves time and may include items overlooked when the scope is created manually. The second area is like the agile concept. Once a scope statement is documented, an AI-based algorithm using NLP reviews the document to look for errors, omissions, and gaps.

One of the discoveries from the work of Professor Flyvbjerg is that projects that start well are less likely to become project disasters. An accurate scope statement supports that finding. Applying AI to ensure an accurate and comprehensive scope statement significantly improves the project methodology.

Project procurement management is critical to the success of many projects. There are three important areas where artificial intelligence (AI) technology will change current practices: the contract statement of work, vendor selection, and tracking procurement progress.

The contract statement of work (SOW) takes on additional significance when outsourcing work to a vendor. The SOW is the basis for requesting bids, evaluating potential vendors, and finalizing a contractual agreement to deliver the SOW contents. Therefore, the SOW needs to be accurate and comprehensive. Natural language processing (NLP) is an essential tool to provide this requirement. NLP uses documents from similar projects and the wording in the SOW to find errors or omissions. NLP performs entity recognition to extract objectives, deliverables, and stakeholders. Sentiment analysis is performed to analyze the tone or content of communication and ensure no issues or conflicts. NLP identifies and extracts keywords critical to project success, then compares that data to find omissions.

The vendor selection process is one of the most challenging areas to maintain ethical behavior. NLP software can remove human bias, similar to current software that scans a candidate’s resume to determine how well they meet job requirements. In addition, seller bid submissions will be automatically checked and ranked based on the selection criteria.

To track vendor progress, the performing organization needs to observe vendor behavior and, if possible, access vendor project data. Robotic process automation (RPA) software creates reports and determines the most critical metrics. A machine learning algorithm detects anomalies and predicts future performance. Unsupervised learning performs clustering of data, compares trends in performance, and highlights variances that may result in unacceptable deviations. This allows both the vendor and project manager to be proactive to avoid any deterioration in the probability of project success.

The future of project procurement will be based on applying AI technology.