AI in Project Management Tools & The 3-Question Test | Sanjana

Introduction: AI Features in Jira, Trello, Asana, and Notion

[0:00]Hi, I'm Sanjana, a project manager at Humanitarian's AI, and this video is about the AI that just showed up inside your project tools Jira, Trello, Asana, Notion. It can write your tickets, summarize your stand-up, even suggest what to do next. So, the real question isn't whether it's smart, it's which of your project work you should actually hand to it and which you should keep. I'll give you a three-question test for exactly

Autopilot vs. Fast Intern: Knowing AI's Limits

[0:25]That. Here's the whole idea in one breath. It's tempting to think the AI in these tools will run the project for you. It won't. What it's genuinely good at is the busy work turning notes into tickets, summarizing, tidying. The judgment stays yours. So, here's the situation. Almost every board you open now has an AI button, and the pitch is always the same. Point it at your project and it'll basically manage it.

The 3-Question Test for Task Delegation

[0:52]But, watch what happens when you actually lean on it. Ask it to run the sprint and it will confidently commit dates it has no way to know. Off a board that's half out of date. It doesn't fail loudly, it fails politely with a clean-looking answer. Treat it like an autopilot, it's a liability. Treat it like a very fast intern, it's a gift. So, before you hand anything to the AI, run it through one test. Three

Practical Walkthrough: Converting Standup Notes into Clean Tickets

[1:16]Questions. One, is it mechanical? Reshaping information that already exists, summarizing, drafting, sorting, or is it a call like choosing what matters? Two, is it reversible? Can you check it in seconds and undo it, or do teammates act on it the moment it's posted? Three, is the truth actually in the tool, or does it live in someone's head, a client call, a Slack thread the AI can't see?

The Autopilot Trap: Committing Dates and Stale Boards

[1:42]Three greens, let the AI draft it. Any red, the AI can suggest, but a human decides. Let's run a real one. Here's a stand-up note, the kind of half-sentence mess we all paste into a board. Watch me hand it to the AI, but only for the mechanical part. Turn this into clean tickets. Don't decide anything. Don't set dates, just draft. And there it is.

The 4-Verb Rule: Draft & Surface vs. Decide & Commit

[2:05]Four messy lines became four clean cards, a bug, a bug, a content task, and ops note, each with a type and a rough size. Now, score it on the test. Mechanical? Yes, it just reshaped what I already wrote. Reversible? Completely, I can read all four in 10 seconds and fix any of them. In the tool? Yes, everything it used was right there in my note. Three greens. This is exactly the work to hand over, and it just saved me the boring 10 minutes. Now, the trap

Hands-On Prompt Exercise: Auditing Your Own Project Board

[2:31]Because AI is helpful on my board is too easy to believe. Same tool, same AI, one step further. Pick this sprint and commit the dates. Score it. Mechanical? No, that's a judgment call about what matters. Reversible? No, the moment it's posted, five people plan around it. In the tool? No, the reason we're doing the export work first is a customer call that never made it onto any card. Three reds. And here's the sneaky one. Even just summarize our progress turns red if the board is stale, because now the AI is confidently reporting fiction. The test is what tells them apart. So, here's the rule you can carry to any tool. Give the AI the two verbs it's built for, draft and surface. Draft the tickets, the summaries, the first pass. Surface what's stale, unowned, or blocked. Keep the two verbs that need a human, decide and commit. What we do this sprint, what date we

Promise, how someone's doing. Run the three question test when you're not sure which side a task is on. That's the whole skill, not use AI or don't, but knowing where the line is. Your turn. Open your own board and paste a real mess into clawed loose notes, half-written cards, a stand-up transcript. This prompt tells it exactly where the line is. Draft the tickets and surface what's stale, but stop at judgment. Don't pick the sprint or commit dates. And this is the important line. For every call it can't make, it has to ask you the one question it would need answered. A good answer comes back with clean drafts and a short list of questions. A bad one quietly makes the decisions for you and now you know that's the exact thing to catch. AI in project management tools with Sanjana Rao for humanitarian's AI. Let it draft and surface. You decide and commit. Thanks for watching.