Skip to content

Decide Open Gaps

A gap is a capability the design needs that your approved technology catalog does not cover yet: the one thing the automated loop refuses to decide alone. Gaps are always internal capability slots. People, external systems, and a legacy system being migrated off are context in the design and never ask for a technology.

Gaps surface in three places, all showing the same state:

  • The Automated Design panel pauses at Needs your input and lists each gap as a decision card.
  • The Assurance tab counts open design gaps.
  • The Review & Approval readiness board blocks approval with one blocker per open gap.

Read a Decision Card

Each card names the capability and offers ranked options. Every option carries four short assessments: capability fit, policy alignment (how it lines up with your applicable regulations), integration fit with the rest of the design, and trade-offs. Two chips matter:

  • existing contract / preferred vendor: choosing this option leverages a commercial agreement you already have (see Vendor Contracts), with the discount shown.
  • fills N gaps: one technology can satisfy several open capabilities; choosing it decides all of them at once.

Make the Decisions

  1. Press Choose on an option (or Choose for all when it covers several gaps). To use something not listed, type it into …or enter a different technology on that card.
  2. The panel header tracks your progress ("3 of 6 selected, not applied yet"). Selections are kept if you navigate to another tab or reload; nothing is lost until you apply or deselect.
  3. Press Apply N Decisions, or the main button, which reads Apply & Re-run while decisions are pending. Either way, pending selections always ride the run; no button starts a run that silently discards what you chose.

Applying does two things, in order: each chosen technology is approved into the technology catalog for that capability, and the design re-runs.

Decisions Are Recorded, Not Suggested

An applied decision is platform state. The approved technology carries the capability it was approved for, and on every later run a still-open slot covered by the catalog is selected deterministically, whatever the AI model does. One decision converges the loop: decide, one re-run, done.

Permissions

Applying decisions writes to the technology catalog, so it needs the manage-catalog permission (architects and org admins have it). Anyone who can view the package can see the pending decisions.

When You Should Not Decide Yet

If the choice belongs to someone else or another quarter, create a defer-technology-choice exception on the Assurance tab with a reason and a review-by date. An active exception covers the gap for approval purposes; it stops covering automatically when it expires. See Exceptions and Deferrals.

You can also resolve a gap manually in the design editor, which locks your choice as an architect override. See Edit and Lock the Design.

After Deciding

The approved technologies appear in Settings → Technology catalog like any other entry: they can be tagged, marked in use, or later forbidden. The re-run design shows the capability as selected with your technology, and the readiness blocker clears on its own.