AI With Human Oversight

HKA's global AI program is known as AiR — a dual reference to Ai Resources and our value proposition: Anticipate, Investigate and Resolve. It is the umbrella brand for all AI initiatives at HKA. HKA has been using Microsoft 365 Copilot for some time and uses Jylo for the benefits it brings for governance, human oversight, quality control, compliance and auditability — from its legal technology origins. Jylo is a secure, scalable AI platform that works with multiple leading AI models to provide AI assistants, collaborative workspaces, document analysis, workflow automation and reusable "playbooks". HKA's review of other AI tools continues.

HKA Tech utilises the HKA corporate tools and, in addition, specific AI tools including ChatGPT/OpenAI and Anthropic's Claude in association with our enhanced Business Intelligence for dashboards and reports. We also use Fellow for the AI, privacy, security and data governance that it brings to our meetings.

HKA Tech's knowledge and expertise in AI is available to our Clients at many levels, as the majority of our clients are facing challenges with AI as it continues its rapid expansion across so many areas of business. We have always been a believer in what has become a current-day mantra of "Promise with Precision". This is particularly important with AI: our objective is to ensure that the AI capabilities we represent to our Clients — whether our own or those of the products and services we provide — clearly separate what is available today from what is planned for the future.

AI-assisted processes continue to improve many areas, from intelligent searches to working with AI agents across cost management, workflows and external systems by MCP.

HKA Tech and AI. HKA Tech's use of AI delivers tangible benefits to our Clients in areas that include:

• Enhanced business intelligence dashboards and reports, with easier ways to interact using natural language and generate additional outputs such as executive summaries; advanced anomaly detection; predictive analysis/forecasting; and improvements in data quality management and automations.
• Enhanced management of meetings held through implementation and ongoing support — with Fellow's AI producing actionable meeting notes, video recordings, documented topics, key actions and decisions, all with configurable security controls.
• Enhanced documentation of the completed implementation, allowing assembly of information to the level required by Clients and in formats that can be readily searched and retrieved — including progressive provision throughout the implementation to assist configuration reviews and as training material.

HKA's ongoing developments in the use of AI include working with multiple leading AI models across everyday operations, workflows and decision making; further developments in the use of AI for business intelligence, such as our own AI Trainer currently working on natural language programming; enhancement options for AI-based production of documentation for Client-specific reference and training material; and developing guidelines around the controlled use of standards and frameworks such as Model Context Protocol (MCP).

PMWEB and AI. PMWEB's latest (2026) version includes a number of specific AI features delivered through PMWEB AI, embedded within the platform and governed by the PMWEB Intelligence Control Framework. PMWEB AI is not a standalone product and does not operate autonomously. It uses large language model technologies from established AI platforms, including OpenAI and Anthropic (Claude), accessed via controlled service integrations to support capabilities such as intelligent document summarisation and an AI assistant for platform navigation and data retrieval.

All AI capabilities operate strictly as advisory tools. PMWEB AI does not execute transactions, approve actions, or make decisions on behalf of users. All decisions remain subject to existing PMWEB workflows, permissions and human approval, enforced through the Intelligence Control Framework. PMWEB AI operates on data already stored within the PMWEB environment and respects the same role-based access controls, permissions and security boundaries that govern all other platform functionality. Where AI services are invoked, data is processed only for the specific user-initiated request and is not used to train, fine tune, or improve external AI models. PMWEB does not expose customer data to public AI training datasets.

The integrity of datasets used within the platform's AI capabilities is maintained through a controlled and customer-specific training approach. The product leverages AI to improve and refine the individual customer's instance and user experience; however, customer data, prompts, outputs and interactions are not used to train or improve AI models for other customers or third parties. To further improve relevance and accuracy, the application presents contextual feedback prompts to end users — for example, confirming whether a response or recommendation met the user's needs. These interactions are used solely to enhance the effectiveness and personalisation of that customer's implementation while maintaining logical separation between customer environments and datasets.

• Proactive Risk Signals — surface automatic notifications and suggested contingency responses when patterns warrant attention.
• AI-Powered Reporting — shift from manual data assembly to conversational reporting and automated report distribution.
• Intelligent Document Processing — turn critical documents into structured, actionable records ready for review.
• AI Data Control — edit, build and update data tables the way your team talks about them, without touching a configuration panel.
• AI Workflow Intelligence — monitor workflow performance, detect bottlenecks in real time and learn the patterns that precede delays.

AI-enhanced dashboards that surface the insights owners and stakeholders need to guide program delivery with confidence. PMWEB's roadmap for AI enhancements in BI reporting plans a shift from manual data assembly to conversational reporting and automated report distribution. In the meantime, PMWEB's reporting allows extensive configuration of dashboards and reports so users can "slice and dice" information into the formats they require — including visualisations — and easily output to Word, Excel, PowerPoint and PDF.

HKA Tech has dedicated Business Intelligence Services providing AI-enhanced dashboards and reports for PMWEB Clients and other Clients seeking more meaningful information from systems across their business. Our BI/AI team is currently working in the following areas — including within tools such as PowerBI and Fabric, using native features such as machine learning and working with other tools to further enhance these capabilities:

• AI Trainer — continually improving the natural language abilities of the AI to more easily access data and produce responses and visualisations.
• Optimising use of no-code and low-code features — automated forecasting with enhanced predictive analysis, decomposition tools, key influencing factors, anomaly detection and machine learning.
• Interrelationships of BI with underlying data quality — including AI automations around data collection, corrections, transformations, access and storage.

AI-driven adoption tools that personalise in-application guidance to accelerate system proficiency. HKA Tech utilises a number of approaches that facilitate user adoption and ongoing use of PMWEB, including Digital Adoption Platforms (DAP). The use of AI in these approaches — separate and apart from AI in DAPs — is described in the Process Improvements section above.

DAPs operate as an overlay to an application, providing features such as walkthroughs to "hold the user's hand" as they are guided through a process; visuals including video at any step to enhance understanding; functions such as hiding items not needed and validating data entry; and various automations. The ability for AI to significantly improve all processing throughout a DAP is extensive.

One of the leading DAPs is WalkMe, which HKA has supplied and implemented for many Clients with consistently positive feedback. WalkMe is a leader in applying AI to a DAP, with the latest release addressing virtually all of the areas where AI can enhance DAP functionality. Read more about WalkMe's AI features. These include how WalkMe can provide benefits back to the AI, giving AI the real-time context; cross-application reach, where WalkMe operates within an enterprise multi-application environment; and the workflow execution needed to deliver results.

There are a number of DAPs available that vary in their levels of functionality and cost, each with varying levels of maturity in terms of both current and planned use of AI. HKA helps separate fact from fiction, focusing on what AI is currently available in each particular product, as well as how that AI enhancement is relevant to each Client's specific needs. Relevance to the specific product to which the DAP is to be applied is also important, as each product has its own features that may already provide some or part of the DAP functionality. Examples of AI-enhanced functionality we review include:

• Is the Natural Language Processing (NLP) advanced, so users can easily express their questions in a way the AI immediately understands?
• What level of anticipation does the AI bring to the process, such as proactively mapping forward options based on steps to date and known use patterns?
• What visual tools are provided by the AI in each process and do they learn what to display (or not display) based on usage?
• How strong are the AI's data validation capabilities — checking rules for data entry, potential errors and potential corrective actions?
• How well does the AI monitor actual use, including indicators of pain points such as consistently incomplete tasks?