Frequently Asked Questions

Pick a plan on the pricing page and sign up — the Get Started button takes you to the signup form. No sales call required to begin. Point us at your existing site (rsync, SSH, mysqldump, or a container image) and we run a shadow build to verify the platform works for your stack before you cut over DNS. Your first diagnostic report runs immediately after the first deploy.

No free tier and no time-boxed trial. Billing is per-minute on the resources you reserve, so the smallest real commitment is a Starter instance — run it for an hour or a year, you pay only for the minutes of CPU, memory, and storage you actually hold. There's no contract to escape and no credit-card-then-cancel dance; spin one up, look at a real diagnostic report against your own app, and resize or cancel whenever.

Start from your app. Starter (1 core / 2 GiB) fits a static site, a small WordPress, or a Laravel app. Pro (2 cores / 8 GiB) is the common WordPress/WooCommerce and mid-size Magento pick. Scale (4 cores / 16 GiB) suits busier Magento and Shopware storefronts. Enterprise (8+ cores) adds multi-app on shared compute, multi-AZ HA, and dedicated nodes. Every plan ships the same diagnostics, backups, isolation, and compliance — the tier picks resource size, not feature set — and you resize in the dashboard later without a rebuild.

Month-to-month, no contract, cancel anytime. Resources meter per-minute and bill per billing-month. Committed 1-year and 3-year pricing (with a discount tied to commitment depth) is offered later — once we have 6 months of customer data and you have 30 days of running workload — but it's opt-in, never a requirement to start.

No cPanel, no shared servers, no noisy neighbors. Each tenant gets its own isolated MySQL, Redis, S3 bucket, KMS key, Kubernetes namespace, and NetworkPolicy on AWS. You deploy via git push or SSH, not a file manager. The real difference is the diagnostic engine: instead of a dashboard of graphs you read yourself, you get structured findings — the slow query named to the line of code, the memory leak traced to the allocation, the deploy regression diffed against the last green state.

We accept rsync, SSH, mysqldump, or container images. We run a shadow build against your existing app first to verify the platform works for your stack. Then DNS cutover. The diagnostic report runs immediately after the first deploy.

It scales with your availability tier. Single-AZ tenants get backup-based restore (~5-30 min RTO). Multi-AZ HA tenants get automatic failover (seconds RTO). Enterprise offers an uptime SLA up to 99.999% with service credits and a named technical contact. Support first-response SLAs are set by the same tier — from 4 hours on standard up to 15 minutes, 24/7, for the highest availability tier.

Yes. We capture 100% of traces — not sampled — plus logs and metrics flowing through one observability pipeline. The diagnostic engine triages continuously: a critical finding opens a ticket automatically. And it triages before it remediates — a crashed process isn't blindly restarted until we've captured the forensics that explain why it crashed.

Single-AZ tenants have ~5-30 min RTO from backup-based restore. Multi-AZ HA tenants have seconds RTO via automatic failover. The diagnostic engine triages before remediating — restarts don't happen until we capture forensics.

Every plan bursts to 3× your reserved CPU and memory, with the first 10% of the month free — so a normal spike costs nothing extra. Need a permanently bigger base? Resize CPU and memory in the dashboard with no rebuild and no downtime. Multi-AZ HA tenants additionally spread load across availability zones. The diagnostic report tells you exactly what drove a spike so you can right-size instead of over-provisioning.

14 days of point-in-time restore on MySQL (weekly XtraBackup fulls + hourly incrementals + continuous binlog shipping). Workspace and S3 media keep hourly snapshots for 14 days with cross-region disaster recovery. Custom retention up to 7 years is available on Enterprise plans.

Yes. A global Cloudflare CDN sits in front of every tenant, and static media is served from S3 through it. Each tenant also gets its own isolated Redis for application and object caching. Deploys are blue/green, so your cache stays warm across a release instead of cold-starting. If you'd rather bring your own CDN, that's supported too.

Yes — the platform layer is ours to keep current. We ship base images per PHP version per app type and patch the OS, language runtime, and web server (nginx, PHP-FPM) on them. Kernel and CVE patching runs on a defined clock — security-critical fixes land within the PCI 30-day window, sooner for actively-exploited CVEs. You own your application code, themes, and plugins; we own everything underneath them.

Yes. On request we deliver a mysqldump of your databases, an rsync of your workspace, and an S3 bucket copy of your media — same formats we accept on inbound migration. GDPR data-portability requests use the same workflow. Service termination triggers a 30-day data retention period before deletion.

Every plan: dashboard ticket UI + email-to-ticket at [email protected]. Pro adds live chat in the dashboard. Scale adds priority queue routing — your tickets surface ahead of standard tickets in triage. Enterprise adds a named technical contact and a dedicated Slack Connect channel once Slack Connect onboarding clears. Response SLAs are set separately by your availability tier.

No — we don't host mailboxes or run an inbox/IMAP service. For transactional email (order confirmations, password resets) your app sends through your own provider — Amazon SES, SendGrid, Postmark, or similar — configured with standard SMTP or API credentials. We keep the hosting platform focused on running your application; deliverability is a specialized service best left to a dedicated email provider.

Custom containers are on the roadmap. At launch you customize via the deployment config (PHP version, extensions, services, resource shape). Power-user Dockerfiles ship after the platform-managed image path stabilizes.

Yes. The MCP server at /api/v1/mcp exposes the same tool surface our internal agents use — deploy, rollback, diagnostics, billing, environment, self-service. Your Claude Code or Cursor agent uses the same APIs we do. Tenant API keys with three scopes: admin, deploy, read-only.

Resource-based, per-minute. You buy reserved CPU and memory capacity ($50/core-month + $25/GiB-month). Burst beyond your reservation is free up to 10% of the month; past that, overage applies at 2× base rate on the peak above base.

No. There are exactly two billing dimensions — compute (reserved CPU and memory, per-minute) and storage. No bandwidth metering, no per-visit fee, no per-request fee, no egress charge. The Cloudflare CDN in front of every tenant absorbs outbound traffic. A traffic surge shows up only as CPU/memory burst, which is free up to 10% of the month — so a viral day doesn't become a surprise bandwidth bill.

Yes, but the model is per-minute on resources consumed — so a high-traffic bot pays for the resources it uses, no more. No session counting. No bot tax.

Storage is a separate billing dimension at three rates: MySQL $1.00/GB-mo (includes XtraBackup + binlog + DR), workspace $0.75/GB-mo (includes snapshots + DR), S3 media $0.25/GB-mo (includes versioning + CRR).

Enterprise tier lets you run multiple applications on shared compute — a WordPress marketing site and a Laravel admin API on the same pool, or a Magento storefront with a separate headless frontend. Each app has its own deploy lifecycle and its own diagnostic stream; the pool sharing comes from one CPU and memory reservation. Smaller plans run one application per instance.

Yes. Budget alerts at 50% / 80% / 100% of an optional spend cap. The diagnostic report shows exactly what's driving cost so you can right-size before crossing the threshold.

Not at launch. Once we have 6 months of customer data and you have 30 days of running workload, we offer 1-year and 3-year commitments with discount tied to commitment depth.

Two paths. Either push to a GitHub repo and we build on push; or SSH into your workspace and run synsmarts deploy. Both paths produce immutable container images, deployed blue/green so traffic only shifts after health and canary metrics pass — a true zero-downtime deploy. Your cache stays warm, sessions survive the cutover, and a failed build auto-reverts before any customer hits it.

WordPress, Magento, Shopware, Laravel, and any PHP application. We ship base images per PHP version per app type. Magento gets composer install, di:compile, and static-content:deploy in the build, not the deploy. Lighter frameworks skip the steps that don't apply. If your stack runs on PHP, we host it.

SSH into your workspace for a real shell, plus the dashboard for a managed view — no cPanel or phpMyAdmin in the middle. Your MySQL is a dedicated, per-tenant instance reachable from your workspace; your files live in the workspace volume and your media in a per-tenant S3 bucket. Agents reach the same surfaces through the MCP server with scoped API keys. Everything is isolated to your tenant — no shared database server, no neighbors on the box.

Structured analysis of what your application is doing — slow queries with the line of code that triggered them, error patterns with stack traces and reproduction conditions, resource bottlenecks with the contributing requests, deploy regressions diffed against the previous green state. Every report has a finding_id you can cite in tickets and a suggested action. Findings are versioned, redacted, and addressable via dashboard, email PDF, API, MCP, or CLI.

The deploy workflow classifies the change. Code-only takes the standard blue/green path. Additive migrations run while the old code serves traffic. Breaking migrations block on customer approval and require a brief maintenance window.

Yes. Abort the rollout via dashboard, API, MCP, or CLI; traffic reverts to the previous version. Rolling back a breaking schema migration requires a data impact assessment — we lay out the tradeoffs and you choose.

Yes. Buy a second instance at the smallest tier. Full production stack, real AWS, real diagnostics. Deploy the same git SHA to both instances to promote.

SAQ D as a Service Provider. We process and transmit cardholder data via PHP-FPM to external processors (Stripe, Braintree, Authorize.net) but never store it. CDE boundary: Cloudflare → NLB → Envoy → nginx → PHP-FPM.

Cloudflare sits in front of every tenant, absorbing volumetric and layer-7 DDoS and filtering requests through its WAF before they reach your app. The origin is locked down so attackers can't bypass that edge: Authenticated Origin Pulls (mTLS) plus NLB IP allowlisting mean the load balancer only accepts traffic that came through Cloudflare. Envoy adds rate limiting at the cluster edge, and per-tenant NetworkPolicies isolate one tenant's blast radius from another's. If you run your own CDN/WAF, the same origin-lockdown pattern supports it.

Yes — managed, auto-provisioned, and auto-renewed at no extra cost. HTTPS terminates at the Cloudflare edge in Full (Strict) mode, so the hop to your origin is encrypted and certificate-validated end to end, never plaintext. Minimum TLS 1.2 (a PCI requirement), with TLS 1.3 preferred. You never touch a cert file or a renewal cron — it's handled.

Yes. DPA template available. Sub-processor register published. Data minimization, retention schedule, breach procedure, DPIA for AI diagnostics — all documented. EU representative designated upon first EU customer onboarding.

SOC 2 Type II is in progress, with Type I as the interim milestone. PCI-DSS SAQ D as a Service Provider is in place at launch. GDPR with DPA, sub-processor register, and DSAR handling is operational. ISO 27001 and HIPAA BAA are on the roadmap as customer demand and audit cycles allow.

Per-tenant MySQL, Redis, S3 bucket, KMS key, namespace, NetworkPolicy. Your code index is physically isolated per tenant — payment data, session tokens, and PII never cross tenant boundaries.

Our diagnostic agents read redacted telemetry to produce findings. Raw payment data and PII are scrubbed at the log-collection layer before reaching storage or any AI pipeline. Customer consent is required for code-level diagnosis: opt-in for EU and enterprise, opt-out for non-EU SMB. We never train models on your code or data.

AWS us-east-1 (Virginia) for the management plane and primary workload cluster. Cross-region disaster recovery to us-west-2. Multi-region is on the roadmap as a premium tier.