Vantage runs entirely on Cloudflare. Is that still the right bet?
First edition · July 18, 2026 · a ~4-minute catch-up
Seven cards, about 30 seconds each. You get the so-what; the full report carries every source link.
EDGEPOINTE · VANTAGE
The headline · our platform
Yes — we're on the right platform. Stay the course
For our exact shape — many-tenant edge sites, edge AI, and edge SQL — Cloudflare remains the price and global-performance leader over Vercel and Netlify. This report is confirmation, not a migration prompt.
The platform is also converging on one model: Cloudflare is folding Pages into Workers, and as of March 2026 Workers reached full feature parity with Pages. The takeaway for us is simple — build new features on Workers-with-assets, not Pages.
Bottom line: staying is the right call. The only shift is a directional one — new work targets Workers, not Pages.
Adopt now · biggest wins
Four high-value, low-effort features to pull in High impact
Workers Cache
A free tiered cache in front of our Workers — on a hit, Cloudflare serves the page without running the Worker or charging CPU. Cheaper, faster published tenant pages; complements KV publishing.
R2 + Images
R2 is $0.015/GB with zero egress — the right home for tenant media once enabled. The Images binding lets a Worker transform photos on the fly: check cache, check R2, resize.
AI Search
Fully managed grounding with per-tenant search instances spun up on demand — one isolated knowledge index per tenant. The clean architecture for our AI Receptionist.
AI Gateway
One control plane in front of every model call: caching, logging, rate limits, retries, fallback. Low effort, high leverage on cost and observability.
Bottom line: Workers Cache is the highest impact-to-effort item — prototype it first.
Free wins · already ours
Headroom and primitives we're not fully using yet Low effort
D1 at 10 GB
Now generally available, 10 GB per database and up to 50,000 databases per account. Our per-tenant isolation model has room to grow — plus a free 40-60% cut in request latency already in effect.
D1 read replication
Read replicas in six regions at no extra charge — lower latency for far-from-home visitors. Needs a code change to the Sessions API.
Queues on Free
Guaranteed delivery, now free up to 10,000 ops/day. A candidate to decouple our Genesis Pipeline stages and buffer lead events.
Roadmap · what's coming
Where Cloudflare is investing next Aligned with us
Pages absorbed into Workers — no forced-migration deadline, but every major feature now ships for Workers first.
Agents Week 2026 — a big agent-infrastructure push: Dynamic Workers, Sandboxes, private networking, and managed Agent Memory. This aligns with Ask Vantage and the AI Receptionist.
Voice on the edge — real-time voice agents over WebSockets with built-in speech-to-text and text-to-speech, relevant to our WebRTC voice work.
D1 replication maturing and a unified AI Gateway API — one endpoint for any model — are the emerging front doors.
Risks · read this one
Dated items that can break a build or start a bill Two quick config checks
workers-types v5 · in effect Jul 3
Dated type entry points removed. Verify we don't pin one — a deploy-breaker if we do. Generate types with wrangler instead.
Service environments · in effect 2026
Removed from Wrangler. Check our CI/CD config for any explicit legacy_env setting.
Workflows step billing · no earlier than Aug 10
New per-step + storage billing. Only bites if we adopt Workflows — count steps first.
DO SQLite storage · started Jan 7
Metered on Paid plans; Free tier is safe.
R2 Data Catalog · 30-day notice pending
Analytics feature only. Core R2 media storage is unaffected.
Net posture: low-to-moderate. Only the two tooling breaks can fail a deploy, and both are five-minute config checks. Every billing item is opt-in.
The close · why Cloudflare
The right home for Vantage, confirmed. Right platform
Cloudflare is the default choice for exactly our profile — multi-tenant edge, edge AI, and D1 — and the consensus recommendation for it.
Performance — Workers start in under a millisecond, sub-50ms across 300-plus locations.
Primitives — the widest set under one roof: Workers, KV, R2, D1, Queues, Images, Workers AI. Rivals lean on third parties for the D1 equivalent.
Cost — free egress plus low compute is a fundamentally cheaper curve at scale — the decisive factor for a many-tenant business.
Bottom line: no migration on the table. Two config audits, then prototype Workers Cache and the R2 media pipeline when we're ready.