Tuesday Brief: An AI tool can only work with what you put in front of it. If the latest decision log, the moved date, or the contract clause isn’t there, the answer comes back sounding right but built on the wrong facts. This week, Kubra Terzioglu returns with the second part of her series, on choosing what the model sees before you ask it anything. Part one is here too, along with a replay that takes the same problem into your Microsoft Project schedules.
New Article: Context Engineering for Project Managers - Give AI the Project, Not Just the Question
You wrote a careful prompt and still got an answer that was polished and slightly off. Maybe it used a date that has since moved, suggested an option the contract rules out, or overlooked a decision the steering committee made last month. In the second part of her series, Kubra Terzioglu, PMP, CAPM, PSM I, traces the problem to the model working without the project in front of it.
Her approach will feel familiar to anyone who has put together a briefing pack. Choose what the AI sees, date and source every item, and when the schedule, the change log, and the steering committee disagree, tell it which one has authority. She revisits risk analysis, executive reporting, and change request analysis with the context each one needs. She closes with five questions to ask before any project data goes into a tool.
Read Article (6 mins)
Resource Article: Prompt Engineering for Project Managers
If you missed part one, start here. Kubra treats a prompt as a requirement and gives it five parts: role, context, task, constraints, and output. Each of her worked examples ends with what to verify before the output reaches a stakeholder. Reading it first makes this week’s piece easier to follow, because context engineering picks up where this article leaves off.
Read Article (8 minutes)
Featured Masterclass: Connect Microsoft Project Desktop to Copilot and AI Agents for a Modern PMO
Microsoft Project files hold some of a PMO’s most useful data: milestones, dependencies, baselines, resource assignments, and the critical path. Most of it stays locked inside individual MPP files. In this August session, Isidro Ávila shows how to make that schedule data usable by AI agents.
What You’ll Learn:
- How a Project MCP parser turns MPP files into secure, agent-ready schedule data
- Practical ways to reduce the manual work behind schedule analysis and status reporting
- How Copilot agents can surface schedule, resource, dependency, and delivery risks
- How historical schedule data supports predictive portfolio analytics
🏅 1 PMI PDU: Business Acumen
Watch Replay (60 mins)
Upcoming Live Events
MPUG hosts live events with leading experts on Wednesdays from 12:00 - 1:00 PM ET. Members get live access and replays.
12/2: Upskilling Project Leadership: Building the Capabilities to Lead What’s Next | Learn More + Register Here
View Our Upcoming Event Schedule
PM Best Practice: Date Every Figure You Hand to AI
Before you paste project data into an AI tool, label each figure with the date it was true and where it came from. Kubra Terzioglu points out that when two sources disagree, a model can answer confidently from the wrong one, so name the source that has authority. Dated, sourced inputs also make the output faster to check, because every claim leads back to something you can open.
Date your data this week!
