RepKit Vision: AI-native infrastructure control plane

Vision

RepKit is an AI-native infrastructure control plane for builders, agencies and small software teams.

You connect the providers you already use — git, domains, DNS, cloud, servers, databases, SMTP and logs — and RepKit helps you plan, launch, monitor and operate applications safely.

RepKit is cloud-agnostic and solution-agnostic. A containerised application should be manageable whether it runs on a VPS, Kubernetes, ECS, bare metal or another target.

The problem

Modern teams stitch together too many infrastructure services.

The result is fragmented ownership, poor visibility, inconsistent setup, manual checklists, weak documentation and expensive mistakes.

Backstage gives teams a developer portal. Terraform gives engineers infrastructure-as-code. Coolify gives self-hosters a deployment layer. Cloud consoles expose provider-specific resources.

RepKit's opportunity is to become the AI operating layer above these systems — focused on application infrastructure, not replacing every tool on day one.

• Domain registrars and DNS providers
• Git repositories and cloud/VPS providers
• Databases, SMTP, CI/CD, logging and secrets
• Deployment targets and monitoring tools

Who we're building for

Initial customers care about speed, clarity, cost and portability.

• Solo founders and indie hackers launching SaaS
• Small software agencies managing multiple client apps
• Small SaaS teams without dedicated DevOps
• Internal IT teams that need a lighter alternative to full platform engineering

RepKit Cloud

A control plane for the full application infrastructure graph.

The core flow: create a project → connect providers → describe application intent → generate an infrastructure plan → review resources → generate deployment files → configure DNS, SMTP and secrets → deploy or guide deployment → monitor status and logs → detect drift → use the AI operator for diagnosis and action.

• Projects, environments, connectors
• Infrastructure graph, desired and current state
• Plans, change sets, operations and execution logs
• Runbooks and AI operator

How RepKit is different

RepKit is not trying to be Backstage, Terraform or Coolify.

Compared with Backstage: RepKit is more action-oriented and infrastructure-operational — not primarily a developer portal or catalog.

Compared with Terraform and OpenTofu: RepKit starts from application and business intent, not low-level infrastructure code.

Compared with Coolify: RepKit is broader than deployment — domains, DNS, SMTP, secrets, provider choices, logs, costs and operations.

First release

The MVP is deliberately narrow and safe.

Kubernetes, ECS and advanced cloud-native provisioning come later. The first deployment target is Docker on VPS.

• Project creation and environments
• GitHub / GitLab connector
• Cloudflare and/or IONOS domain/DNS connector
• One VPS provider connector
• Generic SMTP connector
• Application intent form
• Infrastructure plan generation
• Desired state JSON and deployment checklist
• Generated Docker Compose, Caddy and env examples
• AI operator summary
• Operation log with dry-run execution first
• First deployment target: Docker on VPS

Business model

Phase 1 is hosted RepKit Cloud subscription.

Phase 2 adds paid connector packs, managed execution, compliance reports, private deployment agents and support.

Pricing shown on the homepage is directional — not final live pricing.

• Free — 1 project, limited connectors, dry-run planning
• Pro — £15–29/month — multiple projects, hosted AI operator, connector sync
• Team — £79–199/month — team access, audit logs, multiple environments
• Agency — £199+/month — client workspaces, templates, white-label reports
• Enterprise / on-prem — custom

Go-to-market

We start with builders and agencies. Demo-led selling is critical.

The demo shows: create an app → connect domain and server → add SMTP → generate a plan → see missing DNS records → review the deployment checklist → let AI explain what is wrong.

• Launch a SaaS without DevOps
• Cloudflare + VPS + Postgres + SMTP in one place
• Your infrastructure explained by AI
• Stop losing track of client app infrastructure
• Backstage is too heavy. Terraform is too low-level. RepKit sits in the middle.

What we're honest about

Building a control plane is hard. Here's how we think about the risks.

“This already exists.” — Existing tools solve fragments. RepKit combines provider-agnostic application infrastructure modelling with AI-guided operations.

“Connectors are hard.” — We start with a small connector set and make the connector framework the core asset.

“AI cannot safely manage infrastructure.” — AI does not execute arbitrary commands. It proposes typed operations, dry runs, approvals and logs.

“Terraform already does desired state.” — RepKit owns the application infrastructure graph and can later import or export Terraform/OpenTofu if useful.

“Backstage exists.” — Backstage is a framework for developer portals. RepKit is an AI infrastructure operator focused on planning, launching and operating applications.

12-month direction

A practical sequence — infrastructure first, breadth second.

• Months 1–2 — Project, connectors, intent and plan MVP
• Months 3–4 — Real DNS, SMTP, GitHub and VPS checks
• Months 5–6 — Deployment agent, logs, health checks and backups
• Months 7–9 — Teams, templates, cost comparison, provider recommendations and drift detection
• Months 10–12 — Kubernetes/ECS targets, agency workspaces, on-prem/private agent and paid plans