AI Prompts Become Reusable Workflows, Memory Becomes Editable Product Surface

By DripPublished

The gist

This week, Product and UX work shifted from one-off AI interactions to governed, reusable systems with explicit user control over what the product remembers and reuses.

This week’s developments

Figma Turns AI Prompts Into Governed, Reusable Workflows

Figma’s Weave turns a successful prompt-and-edit sequence into a shareable tool, extending the workflow layer that first showed up in agent-ready execution and then in review and QA. The new step is that the same media-editing process can now be rerun without rebuilding it from scratch, making production more repeatable across teams and formats.

Figma is pairing that with tighter operational control: monthly AI credits and a proprietary model, with 500 credits for Starter, View, Dev, and Collab seats; 3,000 for Professional; 3,500 for Organization; and 4,250 for Enterprise, plus pooled or pay-as-you-go overages under admin caps. It also expanded AI coding agents by adding prototype interactions as context for Code Layers, tightening the loop between design, implementation, and review. Google’s Canvas sketch-to-app workflow points in the same direction, but only that specific workflow is supported by the evidence.

For product and design teams, the progression is practical: the value is moving toward packaging repeatable AI processes, defining the inputs they need, and managing usage limits and review gates. Workflow design and system maintenance are becoming part of the job, not just the output.

How do we turn prompts into reusable, governed workflows?

If you're an individual contributor

  • Your prompt work can become a reusable asset — if you can systemize it.
  • Learn to package prompts, inputs, and review steps; the durable value is in repeatable AI workflows, not one-off outputs.

Sources

If you manage a team

  • Your team’s edge shifts from making assets to governing reusable AI workflows.
  • Coach for prompt design, QA, and usage discipline; build shared patterns so work can be rerun safely across projects.

Sources

If you lead the organization

  • AI workflow reuse is now an operating model issue, not a tool experiment.
  • Invest in governed workflows, credit controls, and review gates; your org needs repeatability before AI sprawl creates waste.

Sources

Memory Becomes a Scoped, User-Editable Product Surface

Between Nov. 4 and Nov. 9, 2026, ChatGPT and Claude both made memory controls more explicit and user-manageable, signaling a shift from hidden account-level settings to scoped, item-level control. ChatGPT added the ability to switch a project’s memory mode after creation, introduced project-only memory so chats cannot reference memories or conversations outside that project, and on Nov. 8 enabled direct deletion from the memory-summary page, including “Delete and turn off memory.” Claude’s Nov. 9 redesign replaced daily summaries with individually editable and deletable Topics, while keeping sensitive categories such as health and beliefs excluded by default unless users opt in.

For product and UX teams, memory is now a first-class interaction surface, not a background system behavior. The design problem is no longer just whether memory exists, but how users consent to it, review it, revise it, and set boundaries over time. One limitation still matters: deleting a memory does not necessarily remove the same information from past chats.

For practitioners, this increases the value of explainability, scoped permissions, and lifecycle transparency. Expect more collaboration with privacy, policy, and AI teams, and more responsibility for designing states users can inspect, constrain, and reverse.

How should teams design governed memory boundaries for different users?

If you're an individual contributor

  • Memory UX is now a product skill, not hidden AI plumbing.
  • Learn to design consent, edit, and delete states; your edge is making AI feel controllable, not magical.

If you manage a team

  • Your team must design memory boundaries, not just smarter recall.
  • Coach for scoped permissions, explainability, and recovery flows; review whether your team can handle privacy tradeoffs.

If you lead the organization

  • Memory is becoming a governed surface, and your org must treat it that way.
  • Invest in privacy-policy-AI alignment and lifecycle controls now; teams that can't prove user control will fall behind.

Sources

Part of these trends

Stay ahead in Product & UX Design

Get the weekly Product & UX Design brief in your inbox — the developments, what they mean by seniority, and what to do next.