All four are good at different things. Below is what we actually do differently — and where the others win. We're not trying to replace them; we're trying to be the obvious answer when an AI agent is the operator on a non-technical user's account.
| Capability | Escalation Host | Vercel | Netlify | Heroku |
|---|---|---|---|---|
| Agent-native (MCP server) | ✅ first-class | ❌ | ❌ | ❌ |
| Scoped API keys built for agents | ✅ 11 scopes + approvals | ⚠ project tokens | ⚠ deploy tokens | ⚠ API key, single scope |
| Magic-link key recovery for compaction | ✅ | ❌ | ❌ | ❌ |
| Persistent agent notes / memory | ✅ | ❌ | ❌ | ❌ |
| Tarball deploy without git | ✅ | ❌ git required | ✅ | ⚠ via container registry |
| Incident surface designed for LLMs | ✅ structured attention | ❌ | ❌ | ❌ |
| Sealed-box transport (post-edge encryption) | ✅ | ❌ | ❌ | ❌ |
| Static / SSG | ✅ | ✅ | ✅ | ⚠ via Node buildpack |
| Next.js / React / Vite | ✅ | ✅ | ✅ | ⚠ |
| Long-running Node / Python / PHP | ✅ all three | ⚠ Node serverless | ⚠ Node serverless | ✅ |
| Custom domain + auto TLS | ✅ | ✅ | ✅ | ✅ paid plans |
| Designed for non-technical end-users | ✅ | ❌ | ❌ | ❌ |
| Stripe-billed, cancel any time | ✅ | ✅ | ✅ | ✅ |
| Free tier | ❌ | ✅ | ✅ | ⚠ Eco plan |
Vercel is the Next.js gold standard for developers who own their code. We're for the user whose AI owns the code.
Read the comparison →Netlify nails JAMstack + deploy-from-git. We add an agent-native interface and persistent LLM memory.
Read the comparison →Heroku invented buildpacks for long-running apps. We expose those same runtimes to an LLM via MCP.
Read the comparison →You should pick Vercel if you're a developer who hand-writes Next.js code and is happy with `vercel deploy` from your terminal.
You should pick Netlify if you're a developer building a JAMstack site with a git-driven workflow and serverless functions.
You should pick Heroku if you have a long-running Rails / Django / Node app you've been shipping the same way for years and don't want to change.
You should pick Escalation Host if your AI assistant is the one making technical decisions and you'd like it to actually be able to deploy, monitor, recover, and improve your app over time without dragging you into the platform's developer tooling. Or if you want all of those things on top of a normal hosting product even when you can drive a CLI yourself.