How to plan a project with AI: scope, tasks, timelines and risks

Organising a school trip, launching a small online shop, running a group dissertation or refitting an office all have something in common: they're projects, with a goal, a deadline, interdependent tasks and things that can go wrong. AI is a good partner for the planning stage — it helps you avoid forgetting tasks, order dependencies and think about risks. What it can't do is know your resources, constraints or the priorities of the people involved, unless you tell it.

#Step 1: Define the scope

The Project Management Institute (PMI) describes a project as a temporary effort with a defined beginning, end and outcome. Start by writing that outcome down:

I want to plan this project: [describe the goal]. Before suggesting anything, ask me what you need to define the scope: what's in and out, deadline, budget, people involved and what "done" looks like. One question at a time.

This step prevents the most common mistake: planning something that isn't clear. At the end, ask for a five-line scope summary and keep it — it's your reference whenever someone wants to add things.

#Step 2: Break down the work

A work breakdown structure splits the project into blocks, and each block into concrete tasks, until every task can be assigned and estimated.

Using the scope above, propose a breakdown into 4–6 work blocks. For each block, list the concrete tasks. Each task should start with a verb and be completable by one person in under a week.

Review the list carefully:

• Are tasks missing that you know about but the AI couldn't — internal approvals, local permits, company sign-offs?
• Are any tasks too big? Split them.
• Is anything out of scope? Remove it.

#Step 3: Dependencies and order

Say which tasks depend on others (can't start until another finishes) and suggest an order. Highlight the sequence of tasks that determines the total project duration.

That chain of linked tasks setting the minimum duration is the critical path. A delay on any of them delays the whole project, so they deserve the closest watch.

#Step 4: Estimate realistically

AI estimates are only a guide — it doesn't know your or your team's availability. Do it like this:

1. You, or each task owner, estimate each task's duration.
2. Ask the AI to build a schedule from those estimates, your holidays and people's real availability ("Ana can only give 10 hours a week").
3. Add buffer — underestimating is the norm. Ask it to flag where a delay would hurt most.

#Step 5: Anticipate risks

For this plan, list the 8 most likely risks. For each, give likelihood (high/medium/low), impact (high/medium/low), an early warning sign and an action to reduce it.

Example output for an event:

#Step 6: Set up tracking

A plan without tracking is just a document. Decide:

• Where the plan lives: a spreadsheet, a kanban board or a project management tool.
• How often it's reviewed: say, 20 minutes every Monday.
• What gets reviewed: finished tasks, delays, active risks and pending decisions.

AI can then help you prepare each review: paste in the current status and ask for a summary of progress, blockers and next steps to send to the team.

#What AI doesn't replace

• Conversations with the people involved. A plan the doers haven't agreed to rarely survives.
• Contextual knowledge: internal rules, real budgets, supplier relationships.
• Decisions: AI offers options; deciding is the team's job.
• Confidentiality: for sensitive projects, use generic names or tools your organisation has approved.

#The bottom line

AI speeds up project planning by helping you ask the right questions, break down the work, see dependencies and anticipate risks. But the plan only works if you feed it real data and review it with the people involved. For your own day-to-day tasks, pair this with our guide to planning your week with AI.