Skip to content

Options

Options answers “what could we do, and what is each option worth?” for the selected run. It compares the options for each decision the run’s checks stand behind, recommends one only when it is a clear winner, and lists the scenarios FACE tested. Open it with Options, under Simulate in the rail (/simulate?tab=swarm).

FACE never acts alone. An option you send goes to a named person for approval on Actionable Twins, and nothing changes until they approve it.

What could we do

The page opens on What could we do, under For: {alert} when the run raised an alert on Why. Each decision the run’s checks stand behind is a chip; the one behind Why’s alert comes first. See options on Why opens this page on that decision. Pick a chip to read its options.

If no check on the run stands behind a decision, the page says No check on this run stands behind a decision yet, so there is nothing to compare.

Read the options

Every option is scored on one measure, stated at the top. For example: Every option is scored on one measure: the share of the temperature readings read that sit past a limit their own row states. Each option card shows:

  • the option, as a lever and a change (for example a temperature excursion moved by −1 °C), or by name (Air expedite, Internal transfer, Do nothing);
  • its rank (#{rank}) and its figure with a 95% range;
  • the evidence it leans on: the effect estimated from the run’s rows with its 95% interval, The baseline: the other options’ effects are measured against it., or Effect not estimated: {reason};
  • who must approve it: One named person must approve it. or Two different named people must approve it. An option marked Advice only: approving it records the decision and runs nothing. changes nothing when approved.

An option whose baseline is missing reads Not measured: needs {needs} and is never ranked. Not measured is never drawn as zero.

A clear winner, or none

FACE recommends one option only when its range clears both the next option’s range and the figure as measured. The card is then marked Recommended, and the header says Recommended: {option} with the reason.

Otherwise the header says No clear winner on this data, with why:

  • The best option’s range overlaps the next one’s.
  • The best option’s range overlaps the figure as measured, so it is not clearly better than leaving things as they are.
  • Fewer than two options could be measured.

A sample that did not read every row says so: A bounded sample: some tables hold more rows than were read.

Send an option for approval

Select Send for approval on an option. FACE compares the options again on the server and puts its own figures on the decision, then queues it for a named approver on Actionable Twins. Not sent: {error} means nothing was queued.

What FACE tested

Under the options, one line counts the scenarios the run tested and how many succeeded, with See them.

The scenarios are a List by default: how each ended, as an icon and a word, its title, its business area and what it found. Graph shows the same scenarios as a flow from what was tested to how it ended. Select a scenario to open its details page (/lab/scenario/<id>).

If the run recorded no scenario, the page says This run recorded no scenario.

New Simulation

New Simulation builds and runs a scenario of your own. Back to the options returns you to the page.

Pick a business area and a scenario

Under Scenario categories, pick a business area and one of its scenarios, or New scenario: name it, pick its business area, then what it tests.

Pick what to test and the news to watch

Under What to test, pick one or more items (New item adds a business measure, such as spoiled pallets per week). Under Market news to watch, pick the terms a coming run searches on the web (New term adds your own). Each run searches only a limited number of terms, and the ones your team added go first.

Set the levers

Under Scenario levers, set the levers for what you picked. Until you pick, each lever keeps its usual value.

Save or run

Save Scenario saves it without simulating. Run Full Simulation simulates it.

The result shows how confident the simulation is, the effects it reasoned from (Why: what this draft reasoned from), any anomalies it detected and its forecast. A forecast needs at least two points; one point is not drawn as a trend.

Send to Actionable Twins hands the result to the approval queue:

You seeMeaning
Queued for TwinsA card was queued. VIEW IN TWINS opens it.
No actions were produced by this runThe server accepted the scenario and named no card, so there is nothing to approve.
The handoff to Twins did not completeNothing was queued. The reason is shown.

Found outside your data

When the run searched the web, what it found is shown here, and only here, under SHADOW CRAWLER TELEMETRY: each finding with its source, Relevance: {relevance} and the query that found it (Query: {query}).

How FACE knows this

How FACE knows this (How each scenario ended, what went to the twins, and the run step by step) is folded at the foot of the page. It holds:

  • how many scenarios ended each way;
  • the hand-off to the twins: Sent to the twins: {sent} · removed from twins analyses: {removed}, or why it is not known;
  • the run’s story: what the analysis proposed and simulated, which verdicts went to the twins, and what the twins do next;
  • the agents re-running on their own after a rule was validated or a scenario changed (Running on their own after a rule or scenario change: {agents}).

RPCs behind this page

  • HypothesisService.CompareOptions, SendToTwins.
  • HypothesisService.InteractiveSimulate, FetchService.GetScenarioGraph.

See the API reference.