Devpilot
Servers

Server Recommendations

Let Devpilot suggest the right cloud provider, region, and instance size for your workload using a short guided wizard.

Server Recommendations

Picking the right cloud provider, region, and instance size is intimidating when you're staring at a grid of names like c6i.xlarge or cpx41. Devpilot's Guided Mode takes a different approach: you answer four simple questions, and Devpilot uses real-time cloud pricing plus an AI model to recommend the best-fit server for your workload — with alternatives you can compare.

Recommendations are a starting point, not a constraint. Everything the wizard picks — provider, region, size, image — you can change in Advanced Mode before you click Deploy.

Where to Find It

Open Servers → Provision Server. The modal opens in Guided Mode by default. Switch to Advanced Mode anytime if you already know what you want.

What Devpilot Asks You

What are you deploying?

Pick the workload type that best describes your project:

  • Web app — A dynamic application that serves HTTP traffic.
  • Static site — HTML/CSS/JS with little or no server-side compute.
  • API service — A backend API without a user-facing website.
  • Database — A dedicated database host.
  • WordPress — WordPress specifically (often panel-friendly).
  • Custom — Anything else.

How much traffic do you expect?

  • Testing / development — Just exploring; minimal load.
  • Small team — Internal tools or a new product with light usage.
  • Production — A live product with real customers.
  • High traffic — Established product with heavy, concurrent load.

Where are your users?

  • Africa, Europe, North America, Asia, or Global.

Devpilot prefers providers and regions close to your audience. For example, picking Africa biases the ranking toward Vultr Johannesburg, AWS Cape Town, and Azure Johannesburg.

What's your budget?

Either a preference (Cheap, Balanced, Performance) or a concrete monthly dollar amount between $5 and $10,000. Devpilot uses this to cap the pool — it won't recommend anything over 110% of your monthly budget, and it tries to use the budget well rather than undershoot it by a wide margin.

What You Get Back

After the wizard finishes, Devpilot returns:

  • A primary recommendation with provider, region, instance type, vCPUs, memory, disk, and estimated monthly cost.
  • Up to three alternatives at different price points or providers.
  • A short, plain-English reasoning note explaining why this server fits your answers — including any budget trade-offs.
  • A pre-selected Ubuntu 24.04 LTS image for the chosen provider and region.

You can:

  • Click Deploy to provision the primary pick immediately.
  • Click an alternative to swap it in as the primary.
  • Click Switch to Advanced Mode to tweak any field before deploying.

How the Recommendation Is Built

Devpilot doesn't just return a static lookup table. Each recommendation combines:

  1. Provider ranking. Devpilot first ranks the candidate providers based on your region and budget preference — for example, cheap + Europe leans toward Hetzner, Vultr, and DigitalOcean, while Africa leans toward Vultr, Hetzner, and AWS.
  2. Live instance data. Devpilot queries each candidate provider's API for current instance types in the region closest to your audience. Pricing is pulled live, not cached indefinitely — so the recommendation reflects today's prices.
  3. Budget-aware filtering. Options above 110% of your monthly budget are dropped. Devpilot keeps a mix of in-budget and slightly cheaper alternatives so the AI has real choices.
  4. AI selection. The filtered pool — with vCPUs, memory, disk, and price — is reviewed by an AI model that picks the best primary choice and up to four alternatives, along with a plain-language reason for the pick.

Recommendations use short-lived caching for both the provider data and the AI response, so repeating the same wizard answers returns quickly without hitting your cloud provider's API again.

AI Provisioning Assistant

Separately from the pre-deploy wizard, Devpilot also ships an AI assistant that helps during and after provisioning. From any provision's detail page, open Ask AI to get plain-English help with:

  • Why a provisioning step failed.
  • What to change (region, instance type, quota, credentials) to get past the error.
  • Whether a retry is likely to succeed or whether something upstream needs fixing.

The assistant grounds its answer in your actual provision — its status, logs, failed steps, and error messages — plus recent successful provisions in the same workspace for comparison.

AI chat is rate-limited to roughly 20 requests per minute per provision, so you won't accidentally burn through quota if you leave a retry loop running.

What Recommendations Don't Cover

  • Compliance-specific regions. If you need a specific region for data residency or compliance reasons (for example, GDPR, HIPAA), switch to Advanced Mode and pick it explicitly.
  • Custom images / AMIs. Recommendations default to Ubuntu 24.04 LTS. Use Advanced Mode to pick a different image.
  • Already-provisioned servers. The recommendation wizard is for new servers. Existing server sizing recommendations are a separate feature coming in a future release.

Next Steps