The decision framing tool (BETA)

The decision framing tool is designed to help people making decisions that improve performance. At this stage it is primarily an "ideating" tool. The process starts by asking you to identify who is making decisions, and then provides several ways to provide background information, from a sentence or two to complete documents in various forms.

The tool helps you frame a decision problem, which consists of identifying performance metrics, types of decisions, and sources of uncertainty that might affect performance. A specially trained AI agent is available to help throughout. You can ask it to provide a rough draft of an entire frame, but we recommend providing as much input as you can. The most important input from you is providing the context and the performance metrics that are most important to you.

After specifying the metrics, you can enter the types of decisions that affect performance. These can be quite broad, such as "Assign drivers to loads" to "Purchase parts from supplier X". You can ask the AI agent to suggest decisions which it considers to have a medium or high impact on at least one performance metric.

If you have any questions, just ? Ask

The framing process is divided into four components, each with its own AI-assists.

1. Problem scope
2. Metrics pyramid tool
3. Decision prioritization tool
4. Uncertainty prioritization tool

Problem scope? Ask

A decision frame reflects the perspective of a decision maker — a person, team, division, or a piece of software. Identify that perspective below, then (optionally) describe the problem itself. From your description, a URL to a case, or an uploaded file, the AI can either read the material so it can inform later steps (metrics, decisions, uncertainties) without generating anything on its own, or produce a rough first draft of the whole framing that you edit and refine below. Treat any draft as an illustration or a starting point, not a finished framing.

Metrics pyramid tool? Ask

Metrics quantify what you want to achieve. They come in three flavors: metrics to be maximized or minimized, along with targets you want to hit, and limits where you specify a minimum or maximum for a metric. The top metric should be in the first category.

Type performance metrics on the left (one per line), then drag each chip into a tier — most important at the top, least important at the bottom. Drag between tiers to re-order, or back to Unassigned to remove.

Decision prioritization tool? Ask

Decisions, which go by many names (including "idea"), represent the ways to impact or influence your metrics. They may be obvious, but they often are not.

List the decisions you'd consider (one per line). The matrix below has one column per tier-assigned metric from the pyramid above, ordered top-to-bottom by tier (left-to-right within the same tier by the order the metrics appear in the metrics list). Click any cell to cycle through H (high impact) → M → L → N (none) → blank. When you're done scoring, drag any row up or down via the ☰ handle to prioritize decisions by their impact on the most important metrics.

Decisions can be general descriptions ("Assigning machines to jobs", "Optimizing warehouses") or specific actions ("Assign machine X to job Y", "Put warehouse in city X"). Use higher levels for general descriptions and lower levels for specific actions. To break a decision down into sub-decisions, click the ▸ button next to that decision (or right-click its row). You can nest sub-decisions to any depth; the metrics pyramid stays fixed.

Uncertainty prioritization tool? Ask

The uncertainties tool works similarly to the decisions tool, except that uncertainties are not nested — all uncertainties live at the top of the framing. Generate ideas is still context-aware: at the top level it suggests categorically distinct sources of uncertainty, and if you have drilled into a subdecision on the decisions side, it suggests uncertainties whose outcomes matter most for that subdecision. Each generated-while-drilled uncertainty is tagged with a green for: subdecision chip in the matrix row below, so you can tell at a glance which uncertainties apply to the whole framing (untagged) versus which were suggested for a specific sub-decision context (tagged). All uncertainties still score against the same metrics in the single matrix. (Click here for a discussion of different categories of uncertainty.)

? Ask