Pricing

Software development cost, in the bands we actually quote

Most engagements land between $15,000 and $120,000. An LLM feature added to a product that already exists starts at $15,000; a full AI SaaS MVP built from nothing tops out around $120,000. The bands below are the ones we actually quote against — not lead-capture placeholders. A discovery call turns the band into one fixed number, in writing, within 48 hours.

How much does each kind of project cost?

Published bands, by what you are actually building. Every figure here is USD, excluding third-party running costs.

AI SaaS product

$40,000 – $120,000

A full AI-powered product: custom pipeline, frontend, backend, and production infrastructure. The span is wide because pipeline complexity dominates — a single-corpus RAG assistant and a multi-agent system with evaluation tooling are not the same build.

8–12 weeks typicalFixed price

SaaS MVP

$40,000 – $90,000

Frontend, backend, database, auth, billing integration, and deployment — a product you can put in front of paying users. Complex data models or AI features push delivery to 14–20 weeks and the price towards the AI band above.

8–12 weeks typicalFixed price

LLM integration

$15,000 – $30,000

Adding an LLM feature to a product that already exists — document Q&A, a chat interface, generation, a code assistant. The variable is not the model, it is how much of your existing codebase has to be touched to reach the data.

2–4 weeksFixed price

NFT marketplace / dApp

Scoped per project

A standard marketplace with minting, listing, and secondary sales takes 10–16 weeks; multi-chain support, royalty enforcement, and advanced search take 16–24 weeks. Contract complexity and audit scope drive both the timeline and the price, so this one is quoted rather than banded.

10–24 weeksAudit extra

Multiplayer game backend

$40,000 – $90,000

$40,000–$90,000, and the number is priced by development time rather than by concurrency directly. The chain is worth stating because it is where most quotes go wrong: the concurrency target decides what has to be built, what has to be built decides how long it takes, and the price follows the time. We have run production infrastructure at 50,000+ concurrent players, and the architecture that holds 500 and the one that holds 50,000 differ in kind rather than in size — which is why the target is the first thing the discovery call establishes.

The monthly infrastructure underneath the build is a separate number, and that one is arithmetic rather than a quote: about $1,200 at 500 peak CCU, $6,900 at 5,000, $48,000 at 50,000 on AWS list prices, line by line with the assumptions.

Priced by dev timeGo · Kubernetes

Dedicated team extension

$4,000–$7,000 per engineer, per month

Senior and mid+ engineers embedded in your team, on your board, in your standups. From $4,000 a month, and $4,000–$7,000 is the band we actually quote — the spread is seniority, not stack: an Unreal engineer and a Go backend engineer of the same level cost the same. Billed monthly rather than fixed-price, because the scope is yours to change week to week. Minimum useful engagement is three months — below that, onboarding eats the value.

MonthlyYour process

What moves the number up or down?

Five things account for most of the spread inside every band above.

Integration count

The single biggest driver. Each external system — ERP, payment provider, identity, a client's legacy database — carries its own auth, its own failure modes, and its own edge cases. Two integrations and eight integrations are different projects at the same feature count.

Whether the data is ready

On AI projects, corpus quality decides the timeline more often than model choice does. Clean, structured, permissioned data means an RAG pipeline in weeks. Scanned PDFs in four languages with no metadata means the data work is the project.

Scale target on day one

Designing for 50,000 concurrent users costs more than designing for 500 — different data partitioning, different failure handling, different infrastructure spend. Building for a scale you will not reach for two years is the most common way to overspend.

Compliance scope

GDPR readiness is included by default on EU-facing work. HIPAA-aligned architecture, SOC 2 evidence collection, or an external smart contract audit are real additional scope and are quoted separately rather than folded into the headline number.

Whether we are starting or inheriting

Taking over an existing codebase can be cheaper than a rewrite, or considerably more expensive. We audit before committing to a scope and give an honest read on which one it is — including when the honest answer is that the rewrite is the cheaper option.

What does not move it

Time zone, team seniority, and IP terms are fixed. Engineers are senior and mid+ only, teams work CET and US Eastern hours, and you own 100% of the code and IP from day one. None of these are levers to negotiate against.

Which engagement model fits?

Three ways to work with us, and the honest case for each.

Managed project delivery

We own the delivery against a fixed price and a fixed scope. Right when you have a defined outcome and would rather buy a result than manage a team. Wrong when the scope is genuinely still moving — you would be paying for change requests instead of software.

Fixed priceWe own delivery

Dedicated team extension

Our engineers, your backlog, your process. Right when you have product direction in-house and need capacity. Wrong when nobody on your side has time to prioritise — an embedded team without an owner drifts, and that is expensive at senior rates.

MonthlyYou own direction

Tech talent acquisition

Recruitment into your own team, permanently. Right when the need is structural rather than a project. Wrong when you need someone shipping this quarter — hiring is slower than either option above, and pretending otherwise helps nobody.

Permanent hireYour payroll

Questions about money

The four that come up on nearly every first call.

Managed project delivery is quoted fixed-price: after a discovery call we scope the work and give a fixed number, so the budget does not move unless the scope does. Dedicated team extension is billed per engineer per month instead, because there the scope is yours to change week to week and a fixed price would be a fiction.
Within 48 hours of the discovery call. The call is free and takes about 30 minutes; what comes back is a written scope with a fixed price against it, not a brochure. If the scope is too vague to price honestly, we say which decisions have to be made first rather than padding the estimate to cover the uncertainty.
Third-party costs you would pay whoever built the product: LLM API usage, cloud hosting, external smart contract audits by firms such as Trail of Bits or OpenZeppelin, app store fees, and licensed art or audio assets. These are passed through at cost rather than marked up, and they are itemised in the estimate so the build price and the running cost are never confused for one another.
Within 1–2 weeks of contract signing. Discovery and scoping usually take the first week of the engagement itself, so a project agreed at the start of a month is typically in its first two-week sprint before the month is out.

Get the number for your project

A free 30-minute discovery call, then a written scope with a fixed price against it within 48 hours. No obligation, and no estimate padded to cover a scope we did not bother to understand.

Message received

We’ll review your enquiry and respond within one business day.

Where the numbers came from