Defining before Designing

Do you ever struggle to gain a clear, unified view of portfolio and project progress?

As teams work in different tools, data and information is scattered across silos, and you can lack the insights you need to make informed decisions.

When working in the paradigm of xPM, the freedom of work is important, and unified reporting ensures you never lose the overview of initiatives’ progress and impact.

Freedom doesn’t have to mean fragmentation.

Working under the xPM paradigm, you use standardized KPIs for unified reporting. No matter how departments or teams work, whether in agile, waterfall, or hybrid environments, unified reporting means consolidating data from multiple tools into a single, clear view. Not only does this provide a better foundation for decision-making – it also leads to happier employees. By allowing people to work in the tools, they prefer they can work more efficiently. It can even improve data quality: if people like the system, they’re more inclined to use it. Unified reporting tools then collects and rolls up the data for reporting and insights.

If you had an AI-agent that knew everything about your project, what would it know?

A successful implementation of AI in your organization is about more than just technology – it’s about process and knowledge.

AI can act as your assistant and give you a shortcut to knowledge and insight – if it knows about your project. With the Microsoft Copilot agent, you can build your own agents that uniquely suit your needs. However, to achieve success, you need to assess what it needs to know.

The 5 questions for successful copilot agent implementation

A lot of companies make the mistake of looking at AI implementation as a purely technological process. But gaining value from Microsoft Copilot is about more than making the technology available. It should fit the organization, and this process begins with asking the right questions. This means identifying who has the relevant knowledge and then interviewing the experienced professionals to understand their work situation.

You should understand the tools, processes, and rhythm: what do they do, how do they do it, and how often? 

The people who know the answers to these questions are your domain experts – not your IT or AI team. That’s where you should start.

This knowledge allows you to design a virtual agent that can fit into this, provide value, and understand the right things.

As an adviser and trainer on Copilot and Agentic AI, I have learned that defining, before designing, is crucial for agent development success.

Here are the five questions, I always ask during implementation.

What would the perfect AI assistant know?

What type of information is relevant for the AI assistant to provide value? This could be

• Project and taskmanagement data 

• Current best practices in the organization 

• industry-specific knowledge 

• domain knowledge 

• Compliance and regulations 

• User preferences & habits 

• News and market trends 

• Historical decision patterns or trend analysis

How would the perfect AI assistant know it?

Mapping out the information requires insight into knowledge sources. Where is the information the AI needs located? How will it access the information?

This could be via 

• Explicit user input

• Behaviour tracking and pattering recognition

• Access to calendar or tasks

• Pretrained models

• Documentation

• Relevant tools or system integration

• Historical data analysis

• And more

When would you use it

Understanding when and how you’ll use your AI assistant helps shape its integration with key systems and the knowledge it needs to be effective. Will it assist during specific times of the day, such as daily stand-ups or end-of-day reporting? Is it most valuable during certain project phases, like planning, execution, or review? Or is its role tied to critical milestones, such as quarterly reviews, stakeholder presentations, or compliance deadlines? Defining these contexts ensures your AI assistant delivers the relevant insights exactly when they’re needed.

Who would use the information from an AI assistant?

Different roles benefit from AI in different ways.

• Project Leads: Use AI insights to track progress, identify risks, and ensure teams stay aligned with project goals.

• Team Members: Employ AI for task management, automated reporting, and quick access to relevant project data.

• PMO (Project Management Office): Gain high-level visibility into project portfolios, resource allocation, and performance trends to support strategic decision-making.

• C-Level Executives: Use AI-driven reports and analytics to assess overall business impact, make data-driven decisions, and prepare for key meetings with stakeholders.

Almost every role in the organization can benefit from AI by getting faster insights, reducing manual efforts, and improving data-driven decision-making. With AI, you can deliver the right information to the right people at the right time.

Where is the AI assistant available?

Defining where the AI assistant should be available will help guide you in your technological investments.  

• Copilot Chat – Engage in real-time conversations, ask questions, and get instant AI-driven insights.

• Microsoft 365 Apps – Access AI assistance within tools like Teams, Outlook, Word, Excel, and PowerPoint for enhanced productivity.

• In-App Integrations – Embedded directly into business applications and project management tools for context-aware support where you need it most.

• Custom Agents: Agents that can handle work flows, and not just generate text but at on your behalf

Embracing Agents for a Smarter Future

In the future, people will work alongside AI. Whether embedded in Microsoft 365 apps, Copilot Chat, or custom business applications, these AI-powered assistants ensure that the right information reaches the right people at the right time. As AI technology evolves, the potential for custom agents will only expand, unlocking new levels of efficiency and innovation. Meet your new digital coworker.