Internal operations platforms, customer portals, dashboards, and marketplaces. Not a brochure website: the thing your staff log into every morning and your customers rely on. We design the data model and permissions before the first screen, build it in visible increments, and keep operating it after launch. You own the code, the data, and the accounts.
A running platform on your own cloud account, an admin interface your staff can operate without a developer, and documentation a different engineer could work from. In practice:
The tested-backup point is not a detail. A great many small businesses have backups they have never restored, which is the same as having none. We restore one before launch so you know the answer rather than hoping.
Web platforms are where the "cheap build, expensive future" pattern is most brutal, because the data model is invisible. You cannot look at a screen and see that customers, branches, and currencies were modelled as an afterthought. You find out two years later, when opening in a second country turns out to require touching every table, and the quote for that change exceeds what the original platform cost.
That is the actual mechanism behind "offshore shops hand you code nobody can maintain". It is rarely that the code is badly typed. It is that nobody decided the shape of the system, so every feature was attached wherever it was quickest, and now the features hold each other hostage. Being paid by the hour for visible screens makes this the rational strategy for the vendor and a disaster for you.
So we spend the first week on the decisions that are expensive to reverse. What is the boundary between one customer's data and another's? Are permissions checked where they can be trusted? Are money and status changes append-only so they can be audited later, or quietly overwritten? Which third-party services can we replace without a rebuild? You get that as a short plain-language document and you are expected to argue with it — it is a business document, not an engineering one.
Then we build in one-to-two-week increments deployed to a real environment, so you are clicking the thing rather than reading status reports, and scope changes are visible while they are still cheap. And we operate what we ship: we run our own products in production, including an AI accounting platform used by Japanese tax firms and an executive dashboard and simulator. Our own shortcuts bill us, which is a very different incentive from a vendor whose involvement ends at handover.
We will not print a figure, because the same-sounding brief can differ by a factor of four. These are the four things that decide it.
One internal team is one application. Staff plus customers plus suppliers plus an admin console is four sets of screens, four permission models, and four sets of edge cases. Cutting a role out of version one is the most effective lever you have.
Payments, refunds, payouts, and escrow bring audit trails, idempotent handling so a retry cannot double charge, reconciliation, and a provider's approval process. Multi-currency settlement adds FX and fee logic on top. Worth doing properly; expensive to do twice.
A documented modern API is days of work. An older on-premise system, a bank gateway, or a platform with a slow sandbox-approval process is frequently the longest pole in the whole project. We identify these in week one, not month three.
Identity verification, personal data rules, financial reporting obligations, or industry-specific compliance change the design rather than adding a checkbox — as they did on the property tokenization platform we built for the Belgian market.
Roughly: a focused internal tool replacing one defined process is typically a 6–10 week build. A customer portal with accounts, permissions, documents, and payments runs longer. A multi-role platform or marketplace with settlement and reporting is a multi-month engagement, and we phase it so a real team is using part of it long before the whole thing is finished.
Replacing the spreadsheet that runs your business — jobs, scheduling, inventory, approvals, quoting, and reporting across branches. Usually the highest-return project an SME can commission, and often paired with an AI layer that reads the incoming documents.
Accounting firms, law firms, and agencies giving clients a private place to submit documents, see status, and pay — which removes the email thread that nobody can audit. Frequently extended later with a mobile app on the same backend.
Buyers, sellers, and an operator in the middle, with listings, payments, settlement, and dispute handling. We have built these across borders, currencies, and languages, including one connecting Thailand and Bangladesh.
A web platform is also the cloud half of most of our other work. The back office behind the POS systems we build is a web application, and so is the admin console behind every mobile app — which is why we prefer to design them together rather than bolt one onto the other later.
The cloud platforms we have architected serve more than 10,000 retail stores and carry over $50M in annual transactions, and our AI systems are in production at more than 200 Japanese firms. Our own products are open for you to click through without registering — the CEO's Cockpit dashboard and simulator is a real product running on fictional data. Two client platforms worth reading in full:
Fractional real-estate ownership with regulated investor onboarding, fiat payment rails, and an on-chain token layer.
Campaign management, local mobile-money rails, and an immutable audit table behind every fund movement.
Stated up front, so nobody wastes a month:
AI agents, document pipelines, and LLM features that remove a named piece of manual work — with a human approval step and an audit trail.
React Native and Flutter apps that ship to both stores from a single codebase, when the app is business logic rather than heavy device work.
Point of sale, RFID/NFC cards, payment terminals, and cloud sync for multi-branch retail, food service, and transport operators.
A website presents information; a web application does work. If people log in, records change, money or documents move, and different roles see different things, you need an application — and the effort sits in the data model and permissions rather than the design. We build applications. If what you actually need is a marketing site, we will tell you that and you will spend far less.
Buy off the shelf whenever it genuinely fits — it is faster and cheaper and we will say so. Custom starts to make sense when your process is your competitive advantage, when you are paying per-seat fees for software that fits three-quarters of the job, when the data has to live inside your own systems, or when you are running the business on a spreadsheet that has already broken once. Bring the spreadsheet to the call; it tells us most of what we need.
All three, from day one. The repository is created under your company, the cloud and database accounts are registered in your company name with you as owner, and the data is yours to export at any time. We are a collaborator you can remove. If you later move to another development partner, you take everything and change nothing — which is the opposite of the arrangement most cheap shops leave behind.
Frequently, and it is a large share of our work. We start with a paid review of one to two weeks: we get it running, read the code, check the database design, and give you a written verdict on what is sound, what is fragile but load-bearing, and what should be replaced. Sometimes continuing is clearly cheaper; sometimes rebuilding one layer buys back years. You get that opinion before committing to a build.
A focused internal tool replacing a defined process is typically a 6 to 10 week build. A customer portal with accounts, permissions, documents, and payments runs longer, usually into a three to five month range. A multi-role platform or marketplace is a multi-month engagement. We work in fixed-scope phases rather than an open hourly meter, so each phase has a knowable budget before it starts.
Hosting sits in your own cloud account, so you are never locked to us by infrastructure. We offer an ongoing arrangement covering monitoring and alerting, backups tested by actually restoring them, security and dependency updates, and a defined response time on defects. Web applications rot quietly — dependencies age, certificates expire, a library gets a vulnerability — and unattended systems are how a working platform becomes an emergency.
That is usually the whole point. We integrate with major accounting platforms, ERP and CRM systems, payment providers, and email, so the platform sits inside your existing process rather than becoming another silo. We also design the API so your own mobile app, or a partner, can use the same backend later — which is why we often build the web platform and the mobile app together.
A 30-minute call with an engineer, in English or Japanese. Show us how the process works today and we will give you an honest read on scope, timeline, and whether you should be building at all.