Custom software, built and run in production.
Platforms, CRMs, client portals, integrations, AI products, desktop apps and bots — designed with you, built by developers working with AI coding agents under review, shipped through dev, uat and prod, and kept running by the same team.
How we ship
- Environments
- dev / uat / prod, a preview per branch
- Builds
- GitHub Actions, never on the production server
- Runs on
- Docker Compose, health checks, daily snapshots
- Consolidated
- 100+ hosted projects onto our own server
- Migrated
- 11 databases to self-hosted Postgres
- Mobile
- Responsive, installable web apps
Platforms we built and run
Described by sector and country, like all our client work, with their real stage.
- Media · South AfricaLive
Creator-marketing platform
Creator profiles, brand campaigns and generated PDF "passports" for creators.
- RestaurantsLive
Booking and customer platform
Bookings, customer records and dashboards for a restaurant group, with the customer base migrated from its previous booking tool.
Read the case study - Training · FranceLive
Management SaaS for training organisations
A management platform for training organisations, in Next.js and Postgres.
- Travel · Southern Africa
Safari quote assistant
A quote assistant with separate dev, uat and prod environments.
Read the case study - Energy · France
Client cockpit
A client cockpit for a wind-energy company.
- Education · Germany
Curriculum hub
A password-protected hub indexing 23 courses from Drive and GitHub for an education institute.
- Own product · France · SpainLive
Directory of sworn translators
4,900+ sworn translators from official lists.
Read the case study - Our own products
Directories, generators, e-learning
A sports directory, document generators and an e-learning platform selling SCORM courses.
Software built around your process
When no product fits how you work, we build one: the screens, the database, the rules and the integrations, in Next.js and Postgres or whatever the job needs, hosted in Docker.
From a prototype to a live product
We start with a clickable prototype on a preview address, cut the scope to what a first customer needs, then ship the MVP through dev, uat and prod. Every branch gets its own preview address, so you test before anything reaches production.
Products with AI inside
Assistants that answer from your documents (we prepared a 1,226-document corpus for one), agents on WhatsApp and the web, generated documents and PDFs — built into a product with its own database, guards and logs. More in AI agents and automation.
Your tools, talking to each other
Integrations we have built: Stripe, transactional email (Resend), Google Calendar booking with Meet links, the Google Ads, Meta Ads and Search Console APIs, the Shopify Admin API, a dropshipping supplier API, NocoDB and GitHub Actions.
Plug your team's Claude into your systems
The Model Context Protocol (MCP) is an open-source standard for connecting AI applications to external systems. We built an MCP server that lets a whole team connect their own Claude to the company's operating system, with every call logged — and we build the same for your tools and data.
Apps for Mac and Windows
Our own video app runs on Mac and Windows and updates itself automatically. For phones we build responsive web apps that can be installed like an app; native iOS and Android apps are something we have not shipped yet.
Bots where your team already talks
Telegram bots, WhatsApp bots and 12 bots on a self-hosted team chat (Mattermost) that run our own company's work.
A CRM that follows your process
We build CRMs around how you sell rather than bending your process to a generic tool: the pipeline stages you use, the fields you need, the reminders and the integrations. We run our own company on one — the CRM pipeline and work system of Takat OS.
Where your customers follow their file
We build portals where each customer logs in and follows their file, their deliverables and their invoices, with access limited to what is theirs — on the same foundations as the client-facing platforms above.
The numbers in one place
Dashboards for bookings and customers, a server health page refreshed every 5 minutes, GA4 analytics and a CRM pipeline view; data extraction and preparation for AI assistants.
Moving without breaking
We moved 11 Supabase databases to a self-hosted Postgres, consolidated 100+ hosted projects onto our own server, moved apps off Cloud Run and Vercel to Docker Compose, moved DNS from Route53 to Cloudflare and automated SSL renewal on 4 AWS servers. The method is in Cloud Run to Docker Compose.
Kept running after launch
Health checks, daily snapshots, automatic rollback, and small changes shipped through the same dev, uat and prod path.
A team that works like ours
Developers who work with AI coding agents (Claude Code) under review — a person reviews and merges every change — on your repositories, with your environments and our shipping rules.
Technical decisions, without a full-time CTO
Architecture, hosting, security, vendors and the roadmap, decided with you by people who run production every day, and written down so your team can follow them.
Screens people understand
Screens designed with you, tested as a working preview rather than a static picture, and checked on a 390 px phone before launch.
From branch to production
- 01
Branch and preview
Every change lives on a branch with its own preview address.
- 02
Build on GitHub Actions
Builds run on GitHub Actions, never on the production server.
- 03
dev, then uat
Changes go to dev, then to uat for you to accept.
- 04
prod
Production in Docker Compose, with health checks and automatic rollback.
- 05
Watch
Daily snapshots, monitoring and a health page.
Related
Frequently asked questions
What kind of software do you build?
Custom platforms, CRMs, client portals, API integrations, AI products, Claude and MCP integrations, desktop apps for Mac and Windows, bots and dashboards — and we host and maintain them.
Do you build mobile apps?
We build responsive web apps that can be installed like an app. We have not shipped native iOS or Android apps yet, and we say so.
How do you avoid breaking production?
Every project has dev, uat and prod environments and a preview per branch; builds run on GitHub Actions, never on the production server; production runs in Docker Compose with health checks, daily snapshots and automatic rollback.
Do you use AI to write code?
Yes: our developers work with AI coding agents (Claude Code) under review, and a person merges every change.
Can you migrate us off our current hosting or database?
Yes. We moved 11 Supabase databases to a self-hosted Postgres, consolidated 100+ hosted projects onto our own server and moved apps off Cloud Run and Vercel.
What should we build for you?
A platform, a CRM, a portal or an integration: tell us how you work and we will send you the plan to build and run it.