How an AI agent can produce a portfolio risk brief with owners and next decisions
This workflow gives an AI agent a defined job, a bounded set of records, and a result a person can review. The agent reads the relevant Asana context, applies the rules in the prompt, and keeps the source behind every recommendation. It returns a proposed handoff rather than taking consequential actions on its own.
Can an AI agent produce a portfolio risk brief with owners and next decisions?
Yes. Start with the scope, date range, decision rules, and fields that identify the right records. The agent can collect the evidence, compare states or sources, mark conflicts and missing data, and organize the result around the outcome above. A reviewer then checks the matches and judgment calls before approving messages, record updates, bookings, purchases, publishing, or other write actions. The guide below shows the records, boundaries, prompt, and handoff needed for this specific workflow.
Define risk in terms of the milestone
Give the agent the portfolio or project list, reporting date, upcoming milestone window, and the team’s risk rules. Useful rules might include a blocked dependency on the critical path, an overdue approval, a milestone with no owner, or a status update older than seven days.
Ask it to trace each risk to the affected milestone. A task being late is evidence, but the brief should explain what cannot start, finish, or be decided because of it.
Example starter prompt
Review these Asana projects: [list]. Evaluate milestones due between [dates] using these risk rules: [rules].
For each risk, cite the project, milestone, task or approval, owner, due date, dependency, and latest relevant status. Explain the delivery consequence and the next decision needed. Separate confirmed blockers, emerging risks, and stale items that need an update. Do not change assignments or dates.Check for stale signals
Completed subtasks do not prove that the parent deliverable is ready. Likewise, a red status from last month may have been resolved elsewhere. The agent should use the most recent task, comment, dependency, and milestone evidence available, then flag contradictions.
Review risks without owners first. An owner recommendation should follow the project’s existing responsibility model, not whoever last commented on the task.
Questions this workflow answers
Which project risks could change a delivery date, and what decision is holding each one up?
An agent can inspect selected projects and trace task-level evidence to upcoming milestones. Define the reporting date, milestone window, and the team’s risk rules: a blocked critical dependency, an overdue approval, a missing owner, a stale status, or remaining work that no longer fits before the due date. The agent applies those rules consistently and names the task, dependency, owner, and latest update behind every finding.
The useful distinction is between lateness and consequence. An overdue housekeeping task may have no effect on delivery. A one-day-late approval may block three teams. The agent should explain what cannot start, finish, ship, or be decided because of the issue. It can group findings as confirmed blocker, emerging risk, or stale signal that needs an update. Contradictory comments and statuses stay visible rather than being resolved from the newest timestamp alone.
Ownership also needs evidence. The last commenter is not necessarily responsible for recovery. Ask the agent to use the project’s existing assignee, milestone owner, responsibility map, or named approver. If those sources disagree or no owner exists, that absence becomes the first decision. Proposed date or assignment changes remain separate from the analysis.
The portfolio lead receives a short summary and a table they can challenge. Each row includes severity, milestone, delivery consequence, source links, decision needed, decision deadline, and current owner. Review the highest-impact items and a sample of “no risk” projects to catch scope or rule errors. On later runs, the agent should compare resolved, new, and worsening risks so the meeting focuses on change instead of rereading the entire portfolio every week.
Expected handoff
The brief should contain a one-paragraph portfolio summary followed by a risk table. Each row needs severity, affected milestone, evidence, consequence, owner, decision date, and recommended next action. Add questions for stale or conflicting records. Approve task changes separately.