Most organisations are exploring AI in project and portfolio management, but very few are truly ready to scale it. To successfully adopt AI, PMOs need three key foundations: a centralised platform, standardised governance, and consistent adoption. Without these in place, AI will amplify existing issues rather than solve them. This article explores the essential building blocks required to prepare your PMO for AI.
What AI (and agentic AI) can already do for PMOs
In Project & Portfolio Management, AI is quickly moving beyond simply answering questions towards agentic capabilities: assistants that can take a goal (e.g., "produce this week's portfolio report" or "re-plan this project after slippage") and then orchestrate the steps across your tools, drafting outputs, prompting for approvals, triggering workflows, and logging actions – all while keeping you in control.
High-value use cases PMOs are adopting now include:
- Drafting and summarising project status updates, highlight reports, and steering packs from structured project data
- Explaining variance by translating schedule/financial/resource movements into clear narrative for stakeholders
- Lessons learned assistance: surfacing relevant lessons from similar past initiatives, highlighting recurring root causes, and recommending actions to avoid repeating mistakes
- Portfolio insights: surfacing hotspots (slipping milestones, overloaded teams, dependency clashes) and answering, "what changed?"
- Planning support: generating work breakdowns, next actions, and schedule options that a PM can refine
Put simply, agentic AI doesn't just generate content—it helps get work done. Microsoft's "Planner agent" is an early example of this direction, although it's still evolving; it's impressive to the point that I've received audible gasps when I've demonstrated it to live audiences.
And in my experience (including Mentimeter pulse-checks from our last few webinars), that's where many organisations are right now: the tools are arriving faster than the understanding, so expectations shoot up, but governance, training and day-to-day adoption lag.
To unlock these capabilities in a way that results in genuine benefit, organisations need the fundamentals in place first: centralised tools, consistent data, clear governance, and strong adoption.
Why PMOs Need Strong Foundations for AI Success
Across multiple AI maturity studies, high-performing organisations consistently demonstrate one thing: they prepare before they deploy.
They invest in the basics, starting with platform consolidation, data consistency, governance, and templates, long before switching on AI tools. It's the classic principle of learning to walk before you run.
As highlighted in our recent AI-focused webinar
- AI success depends on clean, centralised, and consistent data that forms a single source of truth.
- Governance frameworks and standard templates create the structure required for reliable AI insights.
- Only once data and governance are in place can AI deliver predictive reporting, risk identification, and intelligent decision support.
In short: AI is only as good as the environment it operates in and, more importantly, the underlying dataset. Without this, you get artificial information, not the intelligence you’re seeking.
Recent pulse checks from our webinar cohorts reveal three clear trends.
- High M365 adoption, but low consolidation: although 82% of organisations use M365, most teams still operate across multiple non-integrated tools, leading to fragmentation that limits AI readiness.
- Copilot availability is rising: “now or soon” has increased from 50% to the mid-80% range (83–88%).
- Capability understanding lags: only 33% can distinguish Copilot from Copilot agents, with recent cohorts still in the low-40% range.
This highlights a familiar pattern: organisations may have the tools, but the understanding, structures, and readiness needed to unlock real AI value often lag.
Read on to understand the essential steps required to increase your AI maturity.
1. Centralise Tools: Create a Unified Project & Portfolio Management Platform for AI
The problem: Tool sprawl and siloed data
As highlighted in the webinar polls, most PMOs and project teams operate across disconnected spreadsheets, shared drives, project management tools, reporting solutions, and resource management apps.
This is also supported by The State of Project Management 2026 report, which shows that fragmentation remains widespread: 22% of respondents still plan in Excel and 11% report having no project management solution.
In practice, the true scale is likely higher, with 72% spending significant time manually collating reports and around half lacking access to real-time, centralised project KPIs.
This level of fragmentation creates:
- Conflicting data and multiple versions of the truth
- Inconsistent data quality (gaps, duplicates, and stale information)
- High manual effort to produce status updates, dashboards, and KPIs
- Limited automation and AI enablement due to fragmented, non-standardised data
As I often say: “scattered data = missed opportunities”.
The solution: Consolidate into a modern project management ecosystem
Centralisation brings all project data into a single, structured environment. As an award-winning Microsoft Project & Portfolio Management partner, our recommendations typically focus on an M365-based ecosystem consisting of:
- Microsoft Planner Premium (scheduling engine)
- Wellingtone Accelerator+ (the project & portfolio management hub)
- Microsoft Power BI (enterprise reporting solution)
2. Standardise Governance: Ensuring Consistency, Quality and Trust
The problem: Inconsistent ways of working
Without governance, projects are delivered using different templates, varying quality levels, inconsistent RAID processes, and a patchwork of reporting styles.
That low level of maturity is something I still encounter in many of the organisations I speak with. I was in the same boat when I joined the PMO of a FTSE 100 company back in 2005. There was no standardised approach, no common tools, nothing.
Bringing it back to the point, this inconsistency creates issues such as:
- Inconsistent status reporting and “RAG” definitions, making true portfolio health difficult to assess
- Decision-making based on opinion rather than evidence, as assumptions, benefits, and costs are captured differently
- Rework and wasted effort as teams rebuild plans, RAID logs, and reports in different formats for different stakeholders
- Limited automation and unreliable AI insights, as inputs are incomplete, non-standard, or stored in different places
This isn’t just a process problem; it’s a structural one. PMI’s Pulse of the Profession shows that many organisations still rely on a mix of tools rather than a single, consistent PPM operating model.
While 66% of respondents report always or often using project management software, only 32% say the same for portfolio management software, with 75% also relying on budgeting and financial tools such as Excel.
This aligns with findings from The State of Project Management 2026 report, which highlights widespread manual reporting effort and a lack of real-time, centralised KPIs, common symptoms of fragmented tools and inconsistent ways of working.
When teams plan, track, govern, and report in different places and in different ways, governance becomes optional, comparability disappears, and confidence in the numbers drops.
Standardising the lifecycle, templates, and definitions (including what “green” really means), along with stage and quality gates, is what turns project data into something you can trust. It’s also exactly what AI needs if you want insights and automation you can safely act on.
The solution: Standardised templates, processes and controls
To make governance real and usable, treat it as a practical implementation exercise rather than a methodology document.
The approach I always recommend is based on agreeing a “golden thread”: a core dataset and set of standards that make portfolio data comparable (and AI-ready), while still allowing sensible localisation for different departments, teams, and project types.
- Define the golden thread (what must remain consistent). Agree the minimum mandatory standards across the portfolio e.g., risk scoring model, RAG definitions, status cadence, milestone naming conventions, and benefit categories.
- Create a template pack per “project type”. Provide lightweight, role-based templates (Mandate/Brief, Plan, RAID, Status, Closure, Lessons Learned) that are tailored for common project types in each business unit (e.g., IT change, product launch, regulatory, operational improvement).
- Allow controlled localisation. Keep the golden-thread fields locked, but let departments add optional sections or fields (terminology, additional KPIs, delivery artefacts) within the same templates, so you avoid creating shadow processes while still allowing for the capture of additional data.
- Standardise the data model behind the templates. Use consistent field definitions, pick-lists, and IDs (projects, programmes, products, cost centres, resources) so data can be rolled up without manual translation.
- Establish ownership and quality gates. Assign clear data owners (e.g., PMO) and put simple QA checks at key points (initiation, baseline, monthly reporting, closure) to keep information complete and current.
Done well, this improves consistency without forcing every team into a rigid “one size fits all” model, while creating robust, trustworthy inputs that AI can actually use.
3. Support Adoption: Build Trust, Skills & Transparency
Even with the right platforms, data, and governance in place, AI adoption will fall short if people aren’t brought along on the journey. Getting the technical and process foundations right is essential — but so is winning hearts and minds.
You’ll need to address the genuine concerns many users will have (“is this replacing me?”), and build confidence by showing, in practical terms, how AI reduces effort, removes low-value work, and frees people up for higher-value activities.
Below are proven approaches that help organisations move from experimentation to real, sustainable value:
- Start with the “why” (not the features): clearly explain the problems you’re solving — time lost, manual reporting, rework — and what “better” looks like
- Identify PPM Champions: empower people within the business to support users, share success stories, and reinforce new ways of working
- Make training role-based: tailor content for PMs, sponsors, PMO analysts, and delivery leads, grounded in real scenarios (status updates, RAID, reporting, re-planning)
- Teach good judgement, not blind trust: help users validate outputs, challenge assumptions, and improve prompts and inputs
- Be transparent about boundaries: define what AI will and won’t do, including approvals and accountability, so people feel safe using it
- Embed AI into the workflow: integrate prompts, templates, and “next best actions” into everyday processes — not as a separate activity
- Establish feedback loops: capture user feedback early, monitor resistance, and continuously refine guidance and templates
- Reinforce and celebrate success: track adoption, highlight quick wins, and use coaching to prevent teams slipping back into old habits
Conclusion: AI Success Starts With the Basics (and the people)
AI is not a magic overlay — it amplifies what you already have.
If your environment is fragmented or inconsistent, AI will accelerate those problems. But if your foundations are strong, AI becomes a powerful driver of performance.
The good news is that becoming “AI-ready” doesn’t require a complete transformation overnight. It’s about getting the fundamentals right, in the right sequence — and bringing people with you.
In practical terms, that means:
- Centralising tools to create a single source of truth with connected, usable data
- Standardising governance (the “golden thread”) so reporting is consistent and decisions are evidence-based
- Enabling adoption so new ways of working actually stick in day-to-day delivery
A practical starting point is a light-touch maturity assessment (for example, using P3M3) to identify your biggest gaps and prioritise your roadmap.
From there, focus on a single portfolio area or team and pilot a minimum viable standard:
- Agree your core dataset and templates
- Consolidate reporting into one view
- Test, refine, and scale
Depending on your organisation, you may find it easier to standardise processes first (definitions, templates, lifecycle/stage gates) and then centralise tools — or vice versa. The key outcome is the same: consistent data and consistent ways of working.
Once those foundations are in place, these next steps help turn them into real AI value:
- Define success measures for a pilot (time saved, reporting quality, decision speed, stakeholder confidence)
- Put guardrails in place before automation (roles, approvals, lightweight quality checks)
- Turn your pilot into a repeatable rollout kit (templates, examples, FAQs, training), then scale with PPM Champions
At Wellingtone, we support organisations at every stage of this journey, from assessing PPM maturity and AI readiness, through defining the target operating model and “golden thread”, to implementing a Microsoft-based ecosystem built around Planner and Accelerator+, and driving adoption through training, champions, and continuous optimisation
