Figma
Each of these is a real decision Figma is working through, written so you can make the call yourself — incomplete information, someone who pushes back, no answer key.
Figma is a cloud-based, browser-delivered design and prototyping platform that enables design, product, engineering, and marketing teams to collaboratively create user interfaces, wireframes, prototypes, presentations, websites, and design systems in real time. Figma is a public company. Figma builds AI-powered tools used by millions of product teams around the world. Figma has a product called FigJam that includes diagramming capabilities. Figma Make enables teams to go from prompt to production code. The Figma agent is a purpose-built design agent that edits files directly on the Figma canvas. Figma offers a product called Figma Weave in the AI-native creative space. Figma's developer platform includes Figma's MCP Server, REST API, and first-party integrations like Figma for Slack and Figma for ChatGPT/for Claude.
Your first one is free. No card.
46 live right now.
The rest
Shape first-month onboarding before shared files drift apart
New commercial accounts leave kickoff with designers using shared files in incompatible ways. You must weigh a scalable readiness bar against the different ways design organizations actually work.
Settle what evidence a concept needs before it enters shared files
Generated concepts are entering reviews without a shared account of the user need they serve. You have to weigh rapid exploration against evidence strong enough to guide a product decision.
Choose an access rule for Dev Mode before enterprise terms lock
Enterprise administrators are restricting Dev Mode after unexpected charges, and designers are returning to screenshots for engineering handoff. You must weigh predictable access costs against evidence that developers can still build from the canvas.
Choose one change to playback in shared files before design review
Experienced designers are leaving playback for static frames when shared prototypes become dense. You have to weigh speed in a rapid study against enough evidence to change the next interaction decision.
Cluster handoff feedback without losing the evidence behind each theme
PMs are seeing repeated handoff complaints, but the underlying customer evidence is scattered and interpreted differently. You must weigh faster synthesis against the human review needed to keep a cluster trustworthy.
Flag agent edits to shared components before handoff
Design-system leads cannot tell which agent-made component edits received human judgment. You must weigh review confidence against the creative speed of a shared canvas.
Flag partial loads for engineer review without adding alert noise
Researchers are discovering missing product activity after analysis has already started. You must weigh earlier warnings against the review burden of alerts that do not warrant holding data.
Map library readiness before handoff gaps reach engineering
Generated code is exposing inconsistent tokens and missing component states after handoff. You must weigh broad readiness visibility against the limits on enterprise file data.
Prove starter code is reviewable in 70% of mobile flows
Mobile teams are rebuilding approved interface patterns because generated output does not carry their mobile conventions. You have to weigh a fast pilot against the boundary engineers need before generated code reaches a pull request.
Choose an access path before the design-system freeze
The customer has 21 days to approve 30 component changes without unapproved repository access. You must weigh a faster release path against an access model the customer can audit and keep running.
Restore component mappings before the design-system release
Generated code still references retired component properties after the library update. You must weigh a quick configuration change against proof that customer engineers will receive current mappings.
Move guidance lookup into Dev Mode so developers stop asking designers
Developers are reopening libraries and asking designers for guidance that exists but is hard to locate. You have to weigh retrieval usefulness against workspace boundaries and reviewer confidence.
Prove designers will attach intent to 60% of handoffs within one quarter
Pilot teams are losing interaction intent after prototypes enter fast code-generation workflows. You have to weigh a clear developer signal against keeping exploration lightweight for designers.
Trace edit failures while engineers verify every cause
Concurrent edits in large shared files sometimes appear to vanish and return after reconnecting. You have to weigh faster investigation against evidence engineers can defend before changing the multiplayer system.
Surface reviewed states before Dev Mode turns gaps into rework
Developers are reopening shared components because accessibility review coverage is invisible across variants. You must weigh a useful handoff signal against the cost of adding process to the canvas.
Plan Make generations while shared credits disappear
Designers are arriving at prototype review with no shared AI credits left for the generation they need. You have to weigh predictable review capacity against the freedom to explore freely.
Expose component drift before repository pushes become permanent
A release candidate hides shared-component drift behind previews that look correct. You must weigh fast design-to-code handoffs against a review boundary engineers can trust.
Recover component states before review trusts the wrong prototype
Selected and disabled states look right in the generated prototype but behave like their earlier versions in the preview, so a review can approve an interaction that fails on the first click. You have to weigh a fast recovery against making the mapping clear enough for your team to trust.
Rank handoff risk before renewal conversations lose their window
Customer Enablement Managers are rebuilding handoff history across Gainsight and Zendesk while renewal signals become harder to interpret. You must weigh faster preparation against evidence a manager can trust before contacting an account.
Build a handoff programme for Dev Mode despite uneven maturity
Developers open Dev Mode but still return to designers for missing states and token names. You must weigh a repeatable programme against each customer’s different design-system maturity.
Cluster Make feedback without losing the evidence behind it
PMs are reading similar design-to-code complaints as different problems, slowing the path to a usable spec. You have to weigh faster synthesis against the evidence PMs need to trust a theme.
Rank variations in shadow mode until offline evidence reaches 70%
Designers spend a median 8.5 minutes narrowing 12 generated variations before reuse. You must weigh click-derived signals against a ranking model that still respects visual intent.
Decide whether to promote the model despite 11% near-duplicates
Verified review found near-duplicates in 11% of high-volume batches despite a 9% offline preference lift. You must weigh a six-day enterprise commitment against a default model change that is expensive to reverse.
Recover MCP generations while the shared release slot is open
Successful connected asset generations fell to 82% after the prompt-template update. You must weigh a fast context cap against preserving enough file detail for useful output.
Decide what proof backs a generated asset so reviews skip screenshots
Reviewers are reconstructing prompts and inputs from screenshots and approval threads. You need to weigh a useful evidence trail against a first version that stays focused enough to test.
Decide on access for outside agents before the 12-day rollout
Eleven of 27 suspected drift cases are confirmed across 430 external-agent requests. You need to weigh protected brand treatment against keeping approved partner workflows available on time.
Clarify saved variations while this week’s release slot is contested
Forty-nine percent of first-session users leave after generating a variation. You need to weigh a clearer reuse handoff against a simpler way into the workflow.
Choose components as guidance for assets rather than waiting on models
Designers are repeatedly rewriting visual rules and manually comparing each generated variant to approved components. You have to weigh reusable guidance against creative flexibility and clear expectations about model output.
Settle one rubric for focus checks before partners are retrained
Eighteen percent of sampled approved interface states have a keyboard-focus or screen-reader-support gap. You have to weigh a stricter model bar against the useful output designers need to keep moving.
Show every affected asset before a source edit lands
Thirty-nine percent of source edits lead to manual recreation of a connected asset. You have to weigh a faster confirmation flow against the review confidence that keeps approved variants aligned.
Give handoffs evidence before engineering review begins
Engineers are reconstructing component and state intent after generated prototypes arrive for review. You have to decide which evidence earns a place in the first customer-facing review surface.
Set the write-access boundary for Code to Design before Monday's rollout
Eighteen percent of sampled Code to Design updates mapped to the wrong component. You have to weigh the customer's fixed rollout date against drift in shared files that can't be cleanly undone.
Recover first previews while preserving nested components
First-preview success for nested components fell from 84% to 71% overnight. You have to weigh a fast fallback against preserving the component detail designers expect.
Prove the data stays in the region
A rehearsal left 23 derived datasets with unclear regional lineage and four streaming consumers whose routing nobody could confirm. You have to produce proof before the lock date without breaking collaboration reporting.
Decide what moves in the community cutover
A rehearsal found 0.11% of consent records mismatched, 73 threads missing context, and 1,406 identities flagged as duplicates. You get one cutover, so weigh a delay for review against holding the fixed date.
Settle how regulated invoices will work
A signed enterprise deal requires new regulated invoicing before the next renewal, and configuration locks soon. You have to settle tax handling, invoice format, and credits while some tax exceptions are still unverified.
Decide whether to launch with known mismatches
Rehearsal found mismatches in 0.18% of converted files, and automated checks turned up anomalies that reviewers can't yet classify. You have to weigh the committed date against defects that get costly after teams build on them.
Decide whether the key migration is safe to run
Rehearsal left 0.6% of objects with legacy key references, expired recovery envelopes, or paths reviewers couldn't confirm. With no practical rollback, you have to judge when the evidence is strong enough to move customer data.
Decide who gets flagged as a renewal risk
Non-designer participation is down 19% in large workspaces 120 days before renewal, and some of that is deliberate. You have to pick rules that will hold for a quarter, without calling healthy customers at risk.
Design the sharing flow before rollout
Testing before launch showed people mixing up file access, link scope, and outside invitations. You have to make each sharing action clear without slowing the work down.
Decide whether to launch a region-only offering
A readiness review put 92% of file paths inside the boundary, but manual sampling found unclassified ones in cross-region collaboration. You have to decide what evidence is enough before commitments you cannot undo.
Get redaction right before the launch date
The pipeline misses 2.1% of sensitive strings and over-redacts 11% of legitimate UI, and archived packages are costly to correct. You have to weigh protection against the detail that developers still need, on a fixed launch date.
Move history into the new store without losing holds
The rehearsal showed a clash between retention flags and legal holds on some old records, and the residency deadline will not move. You decide on the safeguards, the checks, and the rollback line before deletion starts.
Move the design program without losing approvals
A readiness review found 17 critical files with no clear owner, access, or launch reference, and some of the automated flags turn out to be wrong. You weigh launch speed against the review rigor leadership wants.
Settle terms with the procurement platform
Customers buying through this network close 24 days faster, but the draft agreement asks for broad rights over transaction data and ties your pricing. You decide which terms you can live with for years.