Replit SaaS In Under A Month

Over the last month I have been quietly building something quite serious in Replit: an internal production planning and MES / ERP platform for a multi-machine manufacturing environment.

It is not a simple form-over-database tool. It handles:

• Full job lifecycle from quote through to completion and despatch

• Intelligent job allocation to machines using a weighted scoring model

• Multi-stage production flows with sequence checks

• Cost sheets and estimating with PDF exports

• Digital job packs with document history and cloning

• Operator time tracking and efficiency monitoring

• Machine records, service history and reminders

• Notifications with per-user preferences and job following

• Analytics dashboards for jobs, machines and operators

• Admin reporting with multiple exports and audit trails

• Role-based security, domain-restricted logins, session control and audit logging

In practice it behaves much more like a focused internal product than a basic line-of-business app.

What is going on behind the scenes

Technically, it is a full-stack TypeScript build:

Frontend

• React 18 with TypeScript

• Vite as the build system

• Wouter for routing

• TanStack Query v5 for server state

• React Hook Form with Zod validation

• Shadcn UI and Radix UI

• Tailwind CSS with a custom theme

• Framer Motion for animation

• Recharts for analytics charts

• jsPDF for client-side PDF generation

Backend

• Node.js with Express.js, all in TypeScript

• Drizzle ORM with a schema-first approach

• Neon PostgreSQL as the main database

• Session-based authentication with a PostgreSQL session store

• REST API with Zod request validation

Integrations

• Cloud storage for documents

• Document sync into an enterprise file platform in a UK datacentre

• CRM integration for customer and deal data

• Object storage and email sending on the Replit platform

Code scale today:

• Around 32,000 lines of TypeScript application code

• Roughly 27,000 lines of custom business logic

• Around 5,000 lines in the shared UI component library

• 100+ TypeScript / TSX files

• 20+ database tables with real relationships, indexes and constraints

• 100+ API endpoints

• 40+ React components

This is the size and shape of a genuine SaaS-style product, even though it is used internally.

What it would cost to outsource

If you went to a solid UK software house and asked them to build this from a blank repository, typical numbers look something like this:

• Senior full-stack contractor: about £500–£800 per day

• Small UK agency with proper PM and QA: often £800–£1,200 per day as a blended rate

To reach the level of depth above you are probably looking at:

• Several person-months of backend work

• Several person-months of frontend work

• Architecture, PM, QA and hardening on top

Across the UK market that usually turns into:

• Roughly £120,000 to £200,000 for a solid, pragmatic build

• Potentially £200,000 to £300,000 if wrapped in full enterprise process, UX, documentation and non-functional testing

No one can guarantee the exact quote a specific supplier would give, but these ranges match what similar teams charge for this level of complexity.

The bit people forget: ongoing support

Once live, a system like this does not sit still. You need:

• Bug fixes, security patches and dependency updates

• Small change requests and new reports

• Occasional new workflows as the business changes

Most agencies will price that as a support and change contract. For something of this size, a typical band would be:

• Around £20,000 to £40,000 per year for ongoing support and incremental change, depending on response times and how much new development you ask for.

Over five years, that alone is £100,000 to £200,000 of additional spend if everything is done externally.

Five-year view and likely saving

Put those pieces together:

• One-off build: say £150,000 to £200,000 as a realistic middle of the market

• Five years of support: say £20,000 to £40,000 per year, so another £100,000 to £200,000

Total external spend over five years:

• Roughly £250,000 to £400,000 all in, depending on how “enterprise” you go.

Because this platform has been developed in-house instead of commissioned as a bespoke project, a large part of that external spend has effectively been avoided. Even after you factor in an internal developer’s salary, the likely five-year saving versus a typical UK agency route is comfortably in six-figure territory. A reasonable estimate would be a saving of £150,000 to £250,000 over that period.

For me, the interesting takeaway is not just the feature list, it is the economics. With modern tools, a clear workflow and disciplined iteration, it is now realistic for an operations-heavy business to build its own MES / ERP-grade tools that fit the way it actually works, instead of bending the business around someone else’s product roadmap.