NEARFORM is an Anthropic + AWS + Google + Microsoft Azure partner — a senior-only engineering shop with AI-Native Delivery Pods running Spec-Driven Development. "Open source is the world's largest R&D engine." Cloudflare's developer platform is the natural runtime for the pods themselves: Workers + Workers for Platforms for per-client agent deployments, AI Gateway in front of Anthropic + Cursor, R2 for Spec-Driven artifacts, and Workers AI as the open-source-friendly model layer for the agents your pods build.
The hardest part of selling AI-native engineering is the runtime conversation: where do the agents actually run, how do you avoid vendor lock-in, how do you give clients a foundation that respects their open-source thesis. Cloudflare's developer platform is the only major hyperscaler-class option built around exactly that posture — Workers run on V8 isolates (the engine you already use), R2 has zero-egress S3-compat, and the entire stack works without proprietary SDKs.
Each maps to something published on nearform.com — Spec-Driven Development, the agentic IP, the partner stack, the Forward Deployed Engineering model. The partner-tagged primitives are where Cloudflare's existing partnerships with your partners give you commercial leverage already.
You're an Anthropic Partner. Anthropic + Cursor are both verified on nearform.com apex TXT. AI Gateway sits in front of both: semantic cache, per-client attribution, BYO keys for enterprise clients, budget caps per Pod. The proxy makes your agentic IP cheaper to run and easier to audit for the regulated clients you serve.
Each AI-Native Pod runs agents in a client's environment. Workers for Platforms gives each client their own isolated Worker namespace — your proprietary agentic IP, deployed per client, with per-client keys, egress, audit trails. The "no vendor lock-in" promise is enforced by infrastructure, not by policy.
Your homepage thesis is "open source is the world's largest R&D engine." Workers AI runs open-weight models (Llama, Mistral, DeepSeek, Whisper, FLUX) on Cloudflare GPUs — no proprietary SDK, no lock-in. Pods can choose the model per use case: Anthropic for high-precision reasoning, Workers AI for cost-controlled high-volume inference.
Spec-Driven Development means specs, design decisions, agent prompts, generated code, and evaluation results are all artifacts in a per-engagement archive. R2 stores them at zero egress, with object versioning for full audit history per client engagement. Vectorize indexes them for "find me every spec like this one we shipped before."
FDE means your engineers are embedded in client environments, often touching regulated data (BFSI, Healthcare). Cloudflare Access gives each FDE identity-aware, audit-logged access to client systems — without standing up per-client VPN gateways. The compliance posture sells itself to BFSI procurement teams.
"Agents that survive crashes, hibernate when idle, recall what matters." That's the published spec for production agentic workloads. Durable Objects + Workflows give you the durable-execution runtime — single-writer state per agent session, automatic hibernation, replayable workflows — without standing up Temporal.
You're partners with AWS, Google Cloud, and Azure. Clients keep their data planes on those clouds. Workers + Hyperdrive give Cloudflare-hosted agents pooled, low-latency connections into client data planes on any cloud — no per-cloud SDK, no per-cloud rewrite. The pod ships once, runs everywhere.
The "Inspire Session" on the homepage is a structured senior-led conversation + demo. Demos need to be live, branded, fast — and built quickly without a dev pipeline. Cloudflare Pages is the open-source-native static hosting layer for those demos. Same edge as everything else, no Vercel lock-in.
Sanity is one of your strategic partners. Sanity content is already commonly served through Cloudflare Workers — that's a published Sanity integration pattern. Your Sanity-based delivery work for clients gets edge-resident performance + cache invalidation without per-client CDN configuration.
Senior-only model + AI-Native Delivery Pods means NEARFORM's billable rate is high; the unit economics depend on per-engagement infrastructure cost staying low. AI Gateway + Workers for Platforms turn per-client cloud cost into a per-engagement attributable line — and let you build delivery margin into the pod, not back-of-the-napkin it after the fact.
BFSI clients need SOC 2 + GDPR + DORA. Healthcare needs HIPAA. Retail cares about peak-day uptime. Telecommunications needs deep network integrations. Each industry vertical has different audit cadences and data residency models — and Pods deliver across all four.
Every row sourced from public DNS, nearform.com apex TXT, and HTTP response headers. The orange column is the Cloudflare additive layer that aligns with the partner stack you already have.
You're already an Anthropic Partner. AI Gateway is the lowest-friction observability + cost-control upgrade available on top of an existing Anthropic relationship — no model migration, no prompt rewrite, just a header change. Per-Pod attribution from request one is what makes fixed-fee AI-Native engagements defensible.
Open source is your published thesis. Cloudflare's developer platform is the only major hyperscaler-class option built around exactly that posture — Workers on V8 isolates, R2 with S3-compat API, no proprietary SDKs. The runtime conversation maps your homepage to architecture instead of marketing.
The 5-partner stack already covers everything except the runtime under the pod. Anthropic for the model. AWS/GCP/Azure for the client cloud. Sanity for content. Cursor for engineering velocity. The missing piece is where the agentic IP itself actually runs — the per-client namespace, the inference proxy, the spec artifact store. Cloudflare fills that gap without overlapping any of your existing partners.
The interesting conversation is which of these is closest to your current sprint: AI Gateway behind Anthropic + Cursor, Workers for Platforms for per-client agent deployments, Workers AI for open-weight model serving, or Pages as the open-source-native Inspire Session demo platform.