SerSan · Custom Software & Platforms
Custom Software & PlatformsCustom software built around how your business works.
From internal tools and customer portals to AI-powered products, we design and ship custom software when off-the-shelf tools no longer fit. Management systems, dashboards, web apps, APIs and integrations — built properly, sized to the problem, and handed over as something you own.
2–8 weeks, depending on scope · fixed fee per phase
What goes wrong with business software
Most businesses outgrow their tools long before they replace them.
The spreadsheet becomes a process. The process becomes the business. Then it starts costing hours, causing errors and blocking growth — and the off-the-shelf tool everyone uses can't be bent any further. Custom software is worth building at exactly that point: when the way you work is a real advantage and no product on the market fits it. Sometimes that includes AI. Often the first version doesn't need any.
Typical build includes
Every build is different. These parts keep recurring.
Not every project needs all of it. The mix is agreed before anything is written and sized to the problem — a first internal tool and a customer-facing platform don't carry the same machinery.
- 01
Product engineering: frontend, backend, infra
Web apps, internal tools, portals, dashboards and APIs. Typed services in TypeScript / Python / Go, database design, deploy pipelines. The non-AI 70% of a system that has to be right first.
- 02
AI features, added only where they earn it
Copilots, retrieval, classification and agents, written as typed modules with retry, timeout, cost ceilings and structured logging. And left out entirely when plain software does the job better.
- 03
The AI is tested like the rest of the code
Versioned regression cases for AI behaviour, run in CI alongside the rest of the test suite. Same pipeline that catches a broken endpoint catches a broken prompt.
- 04
Tracing from user click to model output to action
When something breaks you read one record instead of guessing. Every request is traced end to end — submit, backend, retrieval, model call, tool run, response — using OpenTelemetry.
- 05
Handover: documentation, training, optional support
You don't need an internal engineering team. We can own technical delivery and leave you with a system you understand and control. We leave owner and admin documentation, train whoever runs it day to day, brief your developers when you have them, and can stay on for support after launch.
Where this lands
Three moments when off-the-shelf software stops fitting.
01 · Use case
A process has outgrown its spreadsheets
The work happens across spreadsheets, email threads and a tool that was never meant for it. It holds until it has to scale, and every new person makes it worse. We replace it with software shaped around the process.
02 · Use case
A new product, portal or internal system
A customer portal, a booking or management system, an academy or LMS platform, an internal operations tool, a full SaaS product. We design it, build it and ship it, with AI inside only where it makes the product better.
03 · Use case
Extending software you already have
You have a working system and need new functionality, an AI feature, an integration, or simply for it to stop breaking. Tight scope, real integration, no rewrite unless a rewrite is genuinely cheaper.
Deliverables
What you actually hold at the end.
Concrete artefacts, handed over at the end. Yours to keep, extend or give to another supplier: the code, and the reasons behind it.
- Production code in your repository, owned by your team
- Eval harness for any AI feature, run in CI
- Structured tracing, OpenTelemetry-compatible
- Typed AI integration modules (model calls, retrieval, tools)
- Deploy pipeline + monitoring + rollback path
- Runbook, owner training + admin documentation
Selected work
Projects behind this service.
3 projects that exemplify this service — 2 built by SerSan, 1 from prior senior delivery.
Common questions
Answers from real projects.
Do we need an in-house engineering team?
No. Many of our clients have no developers at all. We can own the whole technical side — design, build, deploy, documentation — and hand back a system you understand and control. Where you do have a team, we work alongside them rather than around them.
What about open-source models vs OpenAI / Anthropic?
Both. The model is a choice driven by latency, cost, data residency, and the task, not by ideology. We routinely ship hybrid systems where a small local model handles classification and a frontier model handles reasoning.
Do you build complete products, or only add features?
Both. Some engagements are a whole product from zero — an internal management system, a customer portal, a platform. Others add one feature to software you already run. The scope is set by the problem, not by a minimum project size.
Can you take over a prototype, or software another developer built?
Yes. The Scope phase reviews what's there, what's salvageable, and what should be rebuilt. Usually it's a mix.
Is this just web development?
It's product engineering. Web + mobile + backend + infrastructure + the AI layer that sits inside it. About 70% of a production AI system is non-AI software that has to be right.