Case study · SaaS · Trades
MeisterMeld: voice-first reports and job management for tradespeople
An AI-assisted field-service app that lets trades businesses create work reports by voice and photo right on the job site – instead of typing up notes in the evening.
- Role
- Concept, design, development, deployment, operations (sole developer)
- Status
- MVP live – the first users from the trades are testing the app
- Scope
- 29 database tables (multi-tenant), about 100 API routes, 4 pricing tiers with Stripe subscriptions, 2FA, offline mode and push notifications
- Stack
- Next.js, TypeScript, PostgreSQL, Drizzle ORM, PWA, LLM
- Operations
- Own VPS (1blu), deployment via GitHub Actions
MeisterMeld in 30 seconds
The problem: documentation costs time the job site does not have
In the trades the work happens on site, but the documentation often only afterwards: notes on scraps of paper, photos in the phone, reports at the desk in the evening. That is time nobody can bill – and information gets lost along the way.
The solution: speak first, let the AI structure it
MeisterMeld turns the workflow around. Employees speak the report or upload photos, and an AI (LLM) turns it into a structured work report that the company can review, download as a PDF and attach to the job.
- Voice and photo uploads as input, AI reports as output
- Quick report without a job – when something has to be documented fast, without creating a job first
- Customer, job and employee management in one app
- PWA: installable on the smartphone, no app store needed
Technology and decisions
- Next.js + TypeScript for UI and API, PostgreSQL with Drizzle ORM as a typed data model
- Voice-first and mobile first: the app is built for work on the move, not for the desk
- Own VPS instead of PaaS: full control over data and operations, hosting in Germany
- Deployment via GitHub Actions on every push to the main branch
- Deliberate scope: I do not rebuild accounting and invoicing – specialised programs exist for that. MeisterMeld focuses on what is missing on site: fast, clean documentation.
Lesson from operations: once people work with the app during the day, deployments are no longer an internal matter. Low-downtime releases are part of the product – not just of the infrastructure.
What this project shows
MeisterMeld is the most effort-intensive project in my portfolio because it touches every layer: product idea and UX, AI integration (voice, photo, structured outputs), data model, authentication and roles, deployment and operations with real users.
Keep reading
Case study · Health · Guides
PflegeOasis: care-level calculator, objection-letter generator and AI content with human review
A free portal that helps relatives find their way through the German care system – with calculators, checklists and more than 200 guide articles.
DevOps · CI/CD · Next.js
CI/CD for Next.js 15 on a VPS: GitHub Actions, PM2 reload and Caddy
Manual deployments were slow and error-prone. Today a git push is enough – no downtime and no PaaS.
Case study · Website modernisation · Trades
Kernbohr-Expert: website modernisation with a price calculator and AI readiness
The website of a core-drilling company in Dresden gained a price calculator, the Kerni assistant and a structure that search engines and AI assistants can read well.
AI · LLM · Content pipeline
AI content with human-in-the-loop: 200+ guide articles from LLM drafts
With care law, AI mistakes are not acceptable. This is how I use a language model as a writing aid without handing it the responsibility.
See more or talk
I am looking for a permanent position as a fullstack / AI developer with a DevOps focus and am open to suitable projects.
View MeisterMeld live ↗ Get in touch