Project risk that lives on the board, not in a spreadsheet

Most project management tools treat "risk" as a manual field — a dropdown someone sets once during planning and rarely revisits. Taskive treats risk as a computed, live property of every task. Behind the scenes, an AI scorer (backed by LiteLLM, so the underlying model provider can change without breaking the feature) evaluates each task against a 5x5 impact/likelihood matrix: how bad would it be if this went wrong, and how likely is that given the task's current state.

The output is a single 0-10 risk score rendered as a colored ring directly on the task card — green for low risk, amber for moderate, red for high. There's no separate risk report to open, no spreadsheet to maintain. Risk is simply part of how the board looks, the same way due dates and assignees are.

Impact captures the blast radius of a task going wrong — does it block other work, does it sit on a critical path, does it affect a customer-facing deadline. Likelihood captures how probable a negative outcome currently is, factoring in signals like open blockers, how stale the task has gone without an update, dependency chains, and whether the assignee is already overloaded (drawing on the same underlying signals as burnout detection).

Because scoring runs ambiently — recalculating as task state changes rather than requiring a manual trigger — a task that picks up a new blocker or slips past its due date can shift risk categories without anyone updating a field by hand. This is what "AI is contextual and ambient, not a feature you navigate to" means in practice: the intelligence surfaces where the work already happens.

Risk data isn't locked inside the UI either. It's one of the tools exposed through Taskive's MCP server, so an AI agent connected via Claude or ChatGPT can be asked "what's the riskiest task in this sprint" and get a real answer computed from live data.

Engineering leads use it during sprint planning to catch tasks that look fine on paper but carry hidden risk — a task with three silent blockers, or one assigned to someone already juggling a high-risk load elsewhere. Product managers use it to prioritize review time toward red-ringed tasks instead of reading every card in a stand-up. And because the score updates automatically, nobody has to remember to revisit an assessment they made two weeks ago.

Teams evaluating risk scoring alongside a full platform switch might also compare Taskive against Jira, Linear, or Trello — none of which offer an equivalent ambient risk feature today.