Project management

Meeting Assistance (Transcription, Summary & actions)

AI supports project managers during and after meetings by providing live transcription, structured summaries, and extraction of decisions and action items

In 2026, Microsoft Teams with Copilot is considered the organizational baseline. AI processing and storage must comply with EU data residency requirements.

As we operate within the Belgian public sector, meetings often involve a dynamic mix of Dutch/French (and English) spoken interchangeably. It is therefore essential that AI transcription tools fully support multilingual conversations. Not all tools on the market currently handle this complexity adequately, which can lead to inaccurate transcriptions and lost context. Multilingual support is not a nice-to-have, it’s a critical requirement.

Maturity levels

Level	Name	Description	Technology	Example tools
0	Non-existent	No AI support for transcription or summarization. Manual note-taking only.	none	none
1	Basic assistance	Transcript sporadically available after the meeting. No reliable AI structuring of summaries or actions. Manual consolidation required. Multilingual/code-switching largely unsupported.	Speech-to-text transcription	Basic transcription tools (OpenAI Whisper)
2	Structured assistance	Live transcription, AI-generated summary, and auto-detected action items within Teams. Human validation required before distribution. EU data residency enforced. Multilingual is basic.	Multilingual speech-to-text transcription Integrated LLM-based summarization within collaboration platform	Speech-to-Text (Digital Europe) Microsoft Teams with Copilot
3	Advanced assistance	Improved contextual understanding in multilingual meetings (incl. code-switching). Higher precision in detecting decisions and actions. Structured output aligned with PM practices. Full EU data residency and auditability.	Context-aware LLMs with multilingual optimization and compliance controls	Enhanced Copilot configurations / EU-compliant enterprise AI assistants
4	Near-autonomous	Highly reliable summaries and structured outputs across multilingual contexts. Minimal editing required. Compliance and EU data residency embedded by design. Human oversight remains mandatory.	Advanced enterprise-grade AI assistants with governance-by-design	Mature EU-compliant AI ecosystems

^{AI Maturity Level: Indicates the level the technology vendors claim to have reached in deploying AI solutions that actually work in real-world applications}

IT Software Project Sizing

IT software sizing consists of translating textual requirements (user stories, use cases, specifications) into measurable sizing units such as Function Points (IFPUG, NESMA) or Use Case Points.

In 2026, AI can assist by interpreting unstructured requirements and mapping them to formal sizing elements. AI supports — but does not replace — the human estimator.

The subsequent cost modelling, probabilistic simulation and staffing optimization remain human-driven (e.g., via tools such as Estimals).

All AI processing must comply with EU data residency requirements.

Maturity levels

Level	Name	Description	Technology	Example tools
0	Non-existent	Functional sizing performed manually. Requirements are interpreted and counted by the estimator without AI support.	none	none
1	Textual assistance	General-purpose LLMs used to summarize or structure requirements before manual sizing. No formal mapping to recognized sizing standards. Output requires full reinterpretation by estimator.	General-purpose LLMs ChatGPT / Claude / Gemini	ChatGPT / Claude / Gemini Microsoft 365 Copilot for Excel (AI formulas/summaries)
2	Structured assistance	AI maps textual requirements to structured sizing elements (e.g., EI/EO/EQ, data groups, functional processes). Produces draft counts aligned with a selected sizing method. Human validation mandatory. EU data residency enforced.	NLP	ScopeMaster (greenfield)
3	Advanced assistance	AI understands organizational sizing guidelines, reuse patterns, and boundary definitions via RAG. Supports differentiation between new, reused, and modified functionality. Can assist in classifying relevant non-functional drivers impacting size. Full traceability and auditability required.	Context-aware LLM + RAG to internal sizing guidelines Structured JSON extraction Validation rules aligned with FP standards	Enterprise RAG-based sizing assistants CAST Imaging/AIP for code-based sizing validation (brownfield)
4	Semi-automated	AI performs high-quality draft sizing across large requirement sets with consistency checks and cross-referencing. Highlights ambiguities and missing information. Human approval remains mandatory before formal baseline. EU-compliant governance-by-design.	Tool-using LLM agents Multimodal ingestion (text + code) Automated consistency checking	Agentic EU-hosted LLM frameworks integrated with sizing governance workflows

^{AI Maturity Level: Indicates the level the technology vendors claim to have reached in deploying AI solutions that actually work in real-world applications}

Management product generation

Management product generation involves drafting structured project documents (called Management products in PRINCE2).

In 2026, AI can accelerate drafting, ensure structural consistency, and reuse contextual project knowledge (including indexed meeting transcripts). AI supports the project manager but does not replace accountability or approval responsibility.

Maturity levels

Level	Name	Description	Technology	Example tools
0	Non-existent	Documents created manually. No AI assistance. No structured reuse of meeting transcripts or prior project artefacts.	none	none
1	Ad-hoc assist	General-purpose LLMs used to draft paragraphs via manual prompts. No structured link to project context or official templates. Meeting content reused via manual copy/paste only.	General-purpose LLM prompting	ChatGPT / Claude / Gemini Microsoft Copilot in Word (free prompting)
2	Embedded assist	AI integrated in the authoring workflow. Template sections pre-filled from structured project metadata. First drafts generated consistently. Meeting transcripts indexed per project and used as contextual source. Human validation mandatory. EU data residency enforced.	LLM with template-aware prompting RAG to project workspace (SharePoint/Confluence) Structured section autofill	Microsoft 365 Copilot (Word/SharePoint) Confluence AI Enterprise EU-hosted LLM solutions
3	AI-Human collaboration	AI co-creates full sections with rationale, tracks missing inputs, and enforces house style. Meeting transcripts reused contextually with traceable references. Multilingual outputs normalized. Strong auditability and governance controls required.	Context-aware LLM + memory RAG over project artefacts and policies Structured JSON section generation stakeholder question resolution DLP and redaction hooks	Microsoft Copilot with project context (SharePoint/Teams) Enterprise custom GPTs (EU-hosted) Advanced RAG-based assistants
4	Semi-autonomous	Agentic system proposes complete management products and updates them dynamically when project artefacts evolve (e.g., new meetings, risks, changes). Redlines suggested, not auto-approved. Human approval remains mandatory. EU-compliant governance-by-design.	Tool-using LLM agents Live RAG across M365/Confluence/Jira Workflow automation under policy controls	Copilot Studio agents (EU-configured) Agentic enterprise AI frameworks

^{AI Maturity Level: Indicates the level the technology vendors claim to have reached in deploying AI solutions that actually work in real-world applications}

Project Risk Management

Project risk management covers the identification, assessment, monitoring and reporting of risks throughout the project lifecycle.

In 2026, AI can support early risk detection, consistency in risk descriptions, probability/impact reasoning and cross-project pattern analysis. AI assists the project manager but does not replace ownership, escalation responsibility or governance decisions.

All AI processing must comply with EU data residency requirements.

Maturity levels

Level	Name	Description	Technology	Example tools
0	Non-existent	Risks identified and maintained manually in spreadsheets or static registers. No AI support. Lessons learned are rarely reused structurally.	none	none
1	Ad-hoc assist	General-purpose LLMs used to brainstorm risk lists from prompts or project descriptions. No structured linkage to risk taxonomy or historical data. Outputs require full manual validation.	General-purpose LLM prompting	ChatGPT / Claude / Gemini Microsoft Copilot in Word (free prompting)
2	Embedded assist	AI assists in drafting structured risk statements (cause–event–impact), suggests probability/impact ranges, and checks completeness against predefined risk categories. Human validation mandatory. EU data residency enforced.	LLM + structured extraction Template-aware prompting Rule-based taxonomy validators	Microsoft 365 Copilot (Word/SharePoint) EU-hosted LLM solutions
3	AI-Human collaboration	AI analyses project artefacts (plans, meeting transcripts, sizing outputs) via RAG to detect emerging risks and inconsistencies. Suggests mitigations based on organizational history. Traceability and auditability required.	Context-aware LLM + RAG to project repositories Pattern detection over historical risk registers Structured risk modelling support	Microsoft Copilot with project context (SharePoint/Teams) Enterprise RAG-based assistants
4	Semi-autonomous	Agentic system continuously monitors project artefacts and flags new or evolving risks. Suggests probability updates and mitigation actions. Cross-project pattern mining enabled. Human approval remains mandatory before formal register updates. Full EU-compliant governance-by-design.	Tool-using LLM agents Live RAG across project ecosystems Predictive risk modelling Policy-aware workflow automation	Copilot Studio agents (EU-configured) Agentic enterprise AI frameworks

^{AI Maturity Level: Indicates the level the technology vendors claim to have reached in deploying AI solutions that actually work in real-world applications}