Vercel
Each of these is a real decision Vercel is working through, written so you can make the call yourself — incomplete information, someone who pushes back, no answer key.
Vercel is a cloud platform for building, deploying, and operating modern web and AI-powered applications, with a strong focus on frontend and JavaScript/TypeScript ecosystems such as Next.js, serving individual developers, product teams, and enterprises that need managed infrastructure, security, and compliance for web and agentic apps. Vercel builds the open source Next.js framework, which powers millions of applications. Its compute platform, the layer that runs every workload customers deploy, includes Fluid compute, Vercel Functions, and Sandbox (a primitive for running untrusted and agent-generated code), and Fluid compute is Vercel's own microVM-based compute infrastructure. Its global CDN serves static assets, optimized images, incrementally regenerated pages, and partially prerendered shells to billions of requests for millions of developers. Vercel also builds AI Gateway, which connects applications and agents to hundreds of models with one API key, and eve, a framework for building production-ready AI agents. Vercel Marketplace and Vercel Connect help developers and agents discover, provision, and connect external services to their workloads on Vercel.
Your first one is free. No card.
23 live right now.
The rest
Clarify service access before developers attach it to production
Developers are abandoning service setup when OAuth permissions and environments become unclear. You must weigh a faster connection path against the access details that protect production workloads.
Rank QBR risks when commitments are scattered across account history
QBR preparation is slowed by scattered escalations, usage signals, and unresolved customer commitments. You must decide what AI ranks, what evidence it shows, and where TAM judgment stays required.
Shape a path from v0 prototype to production review
Platform leads are repeating architecture reviews as AI-assisted Next.js applications multiply. You have to weigh a reusable adoption path against the judgment each production application still needs.
Cluster private-access feedback before enterprise requirements disappear
Enterprise connectivity questions are scattered across Linear and GitHub, so recurring adoption blockers look like isolated requests. You must decide what AI groups, what the PM verifies, and how the team keeps evidence attached.
Connect private services without opening a new attack path
Enterprise platform teams are delaying protected workloads because they cannot verify the path from ingress to private services. You have to weigh a narrow, learnable first surface against the demand for broader network control.
Design a kit for Next.js apps while keeping security gates
Approved Next.js applications are waiting on repeated architecture reviews instead of reaching production. You must weigh a reusable path for internal developers against the customer’s need to retain meaningful security and architecture gates.
Commit checkout traffic before the seasonal freeze locks the old route
A small set of checkout responses differs between the legacy and new route during replayed traffic. You must weigh a one-way migration window against evidence that is costly to verify.
Fix the slow pages in App Router without reverting to legacy
The pilot route group is missing the agreed page-load threshold on customer visits. You must weigh a fast performance gain against changes the customer can safely operate after the pilot.
Route compute alerts without letting AI set the incident
Suspicious compute alerts force responders to rebuild workload context while customer deployments wait for shared capacity. You have to weigh a faster route against evidence engineers can trust under on-call pressure.
Prove that storage is clean for agent code without slowing startup
Security reviewers are blocking Persistent Sandbox use until storage isolation is demonstrable. You must weigh a stronger customer guarantee against startup performance and regional consistency.
Trace stale pages while engineers dispute the cache layer
Newly published pages remain stale in some regions while others serve the current version. You have to weigh faster incident triage against evidence strong enough for an engineer to trust.
Show how a release uses the edge without promising a bill
Project owners currently learn about expensive edge behavior after traffic and billing arrive. You have to weigh actionable projections against the risk of showing cost estimates as promises.
Structure renewal calls while sellers keep every commercial promise
Renewal details are getting lost between recorded discovery calls, Outreach threads, and Salesforce records. You have to weigh faster preparation against evidence a seller can verify before speaking to a buyer.
Package a co-sell offer for agencies without custom pricing
Agency partners lack a shared answer when clients compare metered AI-native deployment with flat-rate hosting. You must weigh a repeatable partner package against the flexibility needed in complex customer accounts.
Show what moves before a transfer is confirmed
In testing, 17% of administrators misread at least one asset that would move, and previews missed a permission dependency in 6 of 80 projects. You weigh completion rates against the stronger warnings the risk group wants.
Decide what ships in the new region
The launch date is fixed and audits are done, but the generated compatibility summaries still disagree on some runtime and edge behavior. You decide what ships, what counts as proof, and who is eligible.
Decide how much to redact from build logs
Final checks found seven secret-like strings in minified output, and the detector missed two of them. You have to weigh a missed secret against over-redaction that leaves logs useless for debugging.
Decide what your campaign can claim
The industry window is fixed, and account summaries nobody verified make the audience look bigger than it is. You have to settle the claims, the qualification rules, and the measurement before media spend commits.
Prove the consolidation is ready before the freeze
Earlier moves hit a 19% rate of configuration exceptions, and 14 property owners are still unverified before the seasonal freeze. You have to weigh the fixed launch date against the risk of a move you cannot undo.
Match every route before the one-way cutover
Replayed traffic shows 0.07% of routes resolving differently on the new manifest, mostly nested rewrites, locale prefixes and wildcards. You have to decide what proof is enough before a cutover you cannot undo.
Decide what proof you need before the cutover
Sampled authorization checks disagree 0.07% of the time, mostly on nested group claims and time-bound access, and the cutover cannot be undone. You weigh tighter isolation against locking customers out of previews.
Close the consolidated renewal before peak season
They renew only if their storefronts consolidate under one multi-year agreement; finance wants caps on spend, engineering wants room for uneven releases. You sign terms you cannot walk back before their peak planning cutoff.