From $6,000 · 4–6 weeks
MVP development
Idea to a working product real users can sign up for. Starts with a $500 discovery that produces a real specification and a fixed quote.
How MVP builds work →
Software engineering practice
Ship in six weeks. Built to last longer.
Fifteen years of production architecture, applied to products that need to exist now. We build MVPs, run ongoing development, and stabilize software that grew faster than its foundations — using AI to move quickly, and judgment to make sure what ships holds up.
MVP · Development · Support · Stabilization
Evidence
Systems where the numbers mattered.
0%
Infrastructure cost reduction on a financial media platform re-architecture
0
Visitors an hour served by that platform at peak
0
Insurance policies sold monthly through applications we built
0
Architecting and delivering production software, including teams of 20+
Client engagements are under NDA. Figures are from named work, described by sector and scale only.
The problem
Most companies can't justify an engineering team. They still need engineering.
Hiring a developer is a six-figure commitment and a three-month search. An agency wants a large contract and staffs it with junior developers and a project manager. Freelancers disappear mid-task.
What's usually needed is smaller than that, and better: someone senior who can decide what to build, build it, and still be there next month.
1
The number of people between you and the person writing your code.
Services
Five ways to work together.
From $2,200 / month
Ongoing development
Reserved days each week committed to your roadmap. Planned delivery rather than reactive support.
How ongoing development works →From $250 / month
Support plans
Monitoring, backups, security patches and small fixes, priced as an operating cost rather than a project.
How support works →Review $500 · from $4,500
AI-built app stabilization
Built with Lovable, Replit, Bolt, Cursor or Claude Code, and now it has real users? Security gaps closed, deployments made safe, monitoring and tests added.
How stabilization works →Also available: fractional CTO, from $2,000/month.
Proof
Where those numbers came from.
Financial media platform · India
A legacy PHP system serving market data to retail traders, re-architected onto microservices. Infrastructure costs fell roughly 76% and the rebuild paid for itself inside eighteen months.
Insurance group · Pan-African
Application flows for two insurance products with KYC and risk assessment, serverless on AWS. Around 2,000 policies sold monthly.
Consumer application · production traffic
Manual deploys that took the site down, replaced with load-balanced infrastructure, an automated pipeline and auto-recovery. Releases stopped causing downtime.
Full engagement detail, domain experience and the technical background are on the about page.
Indicative pricing
Indicative, so you can tell if we’re in your range.
| Engagement | Price |
|---|---|
| MVP discovery Full specification, data model, scope line and a fixed build quote. Yours to keep either way. | $500 |
| MVP build Set by scope and complexity. Final price after discovery. | from $6,000 |
| Ongoing development Reserved days each week for roadmap delivery. Includes maintenance. | $2,200–6,000/mo |
| Support plan Monitoring, backups, patches and small changes. Three plans. | $250–1,200/mo |
| Technical review Written report, risk list and remediation plan for an existing system. | $500 |
| Stabilization Set by scope. Final price after the review. | from $4,500 |
| Fractional CTO One or two days a week. Three-month minimum. | $2,000–3,800/mo |
These are real numbers from real engagements, not anchors — enough to tell you whether this is worth a conversation. The final figure comes after we’ve talked, because scope, complexity and what already exists all move it, and quoting before understanding those would just be guessing at your expense.
50% deposit to begin, balance net 14. Discovery and review fees are credited against the work that follows if you proceed within 30 days.
Common questions
Answers before the call.
Why can't I just build it myself with Claude or Cursor?
Sometimes you can, and sometimes you should. If you're technical, still testing whether anyone wants the product, or building something that's mostly forms and a database, go build it. What the tools don't do is decide. The data model that doesn't need tearing up in month three, the permission boundary nobody thought to check, the payment retry that charges twice, and the discipline to leave features out — those are judgment calls. There's a fuller answer on the MVP page.
How much does it cost to build an MVP?
Ours start at $6,000 for a four-to-six week build, with the exact figure set by scope and complexity after a $500 discovery. Well below US agency rates because there is no sales layer and no project management overhead.
Do you use AI to write the code?
Yes, extensively and deliberately — that's what makes a six-week MVP possible. The difference between that and a vibe-coded app is that every architectural decision is made by someone with fifteen years of experience, and everything ships with tests.
How big is your team?
Deliberately small, with a limited number of engagements running at a time. That's what keeps every engagement architect-led rather than handed to whoever is on the bench. If we don't have capacity for your timeline, we'll say so rather than stretch.
Do you bill by the hour?
No. Everything is a fixed price agreed before work starts, set by what's being built rather than how long it takes. You should never have to audit a timesheet to know whether you got value, and we shouldn't be penalised for being fast. If scope changes, that's a change order with its own price, agreed in advance.
Who owns the code?
You do, entirely, from the first commit. It lives in your repository and deploys to your cloud account. No proprietary frameworks, no lock-in.
Next step
Tell us what you're trying to build.
Twenty minutes. You describe the problem, we'll tell you what it realistically takes and what it costs. No pitch — and if we're not the right fit, we'll say so on the call.