LeavePilot, a leave management tool born from a real problem
First an internal tool for a single company with multiple sites. Then a platform hosting different companies on shared infrastructure. The transition between the two is the interesting part.
LeavePilot is a web platform for managing employee holidays, hourly leave and remote work, with a shared calendar, approval workflows, department management, payroll exports and AI-generated reporting. It is free for up to five employees and is built by MAXYMIZE.
Visit the LeavePilot website Features, plans and a free tier for up to five employees.
The problem it solves
Absence management in small and mid-sized companies almost always lives in two places: a spreadsheet only one person understands, and a chain of messages nobody can find when it matters. It works as long as that person is around and nobody disputes a balance.
Full HR suites solve the problem but bring modules, licences and complexity designed for far larger organisations. A company with fifteen employees ends up paying for a system built for five hundred, and using ten per cent of it.
LeavePilot occupies the space in between: it does one thing completely, from a request made on a phone to the export your payroll accountant needs, without demanding weeks of configuration before it is useful.
The decisions that mattered
The project started for one company and became a platform hosting many. That transition forces choices best made at the beginning.
Data isolation as a structural constraint
In a platform hosting different companies on one database, the worst failure is not downtime: it is one company seeing another company data. Organisation membership is not a filter applied wherever someone remembers to add it, it is a condition built into data access. A badly written query has to return nothing, never somebody else records.
The calendar is the interface, not a summary
The product decision is that you request a day off by clicking that day, not by filling a form with two date fields. It costs more to build: multi-day selection, overlap between colleagues, per-department views, and all of it working on a small screen. But that is the difference between a tool people use and a tool people avoid.
Hourly leave, not just whole days
Modelling absence as whole days is simpler and wrong: someone leaving for a two-hour medical appointment has not taken a day off. The data model handles fractions of a day from the foundations up, because adding them later means rewriting every balance, every total and every export already running in production.
A serverless database for irregular load
Leave management traffic is spiky: it concentrates in the weeks before summer and at year end, and is nearly idle the rest of the time. A serverless database matches that profile and costs in proportion to real usage, which is what makes a free tier sustainable rather than a loss.
AI writes the summary, it does not make the decision
Summary reports are generated with a language model, but the boundary is explicit: the AI does not approve requests, does not compute balances and never touches the numbers. It receives already-calculated data and narrates it. Where decisions affect people employment terms, the calculation stays deterministic and auditable.
What it is built with
Technologies chosen to handle the spiky load this domain produces and to keep running costs proportional to usage.
Interface
- Next.js 15
- TypeScript
- Tailwind CSS
- shadcn/ui
- FullCalendar
Data
- Serverless PostgreSQL (Neon)
- Drizzle ORM
- Versioned migrations
Infrastructure
- Serverless functions
- Blob storage
- Continuous deployment
Advanced features
- Generative AI reporting
- Italian, English and Spanish
- Payroll exports
What this means for your project
If you are planning a platform that has to serve multiple customers, these are the knots we have already untied.
- Multi-tenant architectures where data isolation is guaranteed by structure rather than by the discipline of whoever writes the next query.
- The path from internal tool to sellable product, which is almost always the cheapest way to validate a SaaS: solve a real problem for someone first, generalise second.
- Pricing models with a sustainable free tier, where infrastructure costs scale with usage so the numbers still work with many non-paying users.
- Language models embedded in operational processes, with an explicit boundary between what the AI may write and what must remain verifiable calculation.
Frequently asked questions
What is LeavePilot?
LeavePilot is a web platform for managing employee holidays, hourly leave and remote work: calendar-based requests, approvals, automatic balances, department management, payroll exports and AI-generated reports. It is built by MAXYMIZE and free for up to five employees.
Who is it built for?
Small and mid-sized companies, typically five to two hundred employees, currently managing absence through spreadsheets and messages and with no need for the complexity of a full HR suite. It supports multiple sites and departments within the same organisation.
Is my company data separated from other companies?
Yes. Organisation membership is a structural condition of data access rather than a filter applied case by case: a request that does not correctly identify the company returns an empty result, never another organisation records.
Can you build a SaaS platform for my idea?
Yes. A typical path starts with a working minimum version in four to eight weeks, enough to validate the idea with real users before investing in the rest. Our how we work page covers engagement models, timelines and indicative price ranges.
Have a platform in mind?
The cheapest way to find out whether an idea holds up is to build its minimum version and put it in front of real users. Let us talk it through.