Every few weeks an enquiry lands that opens the same way: “We were quoted €40,000 for this, but surely with AI it should be a fraction of that by now?”
It is a fair question, and the honest answer is an uncomfortable one. AI has genuinely transformed how software gets written. It has not, so far, transformed what software costs. Here is what the Irish market actually charges in 2026, what sits inside those numbers, and why the discount most buyers are expecting has not arrived.
TL;DR
- Irish agency day rates in 2026 sit around €600 to €1,000. Freelancers charge €300 to €600, and the national IT contractor average is roughly €572 per day.
- A professional SME website runs €4,000 to €8,000. Standard business sites with integrations reach €6,000 to €22,000, ecommerce €6,000 to €25,000, and custom platforms start around €40,000.
- AI has not reduced project cost. First-year costs with AI coding tools run roughly 12% higher once code review overhead, testing burden and code churn are counted properly.
- Writing code was never the majority of the bill. Discovery, architecture, review, testing, deployment and maintenance are, and AI barely touches most of them.
- Budget for total cost of ownership, not build cost. Plan on 15% to 25% of the build price every year to keep the thing alive.
What the Irish market actually charges
Rates first, because everything else is derived from them. In 2026, Irish agency day rates typically fall between €600 and €1,000. Independent freelancers sit at €300 to €600. Hourly billing ranges from €80 to €200 depending on seniority and specialism.
The contract market gives a useful sanity check. IT remains the highest-paying contracting sector in Ireland, with an average day rate around €572 and a median of roughly €500. Senior engineers command €600 to €900, and specialists in React, Next.js and AI engineering push past €1,000. An agency rate that looks high next to those figures is covering design, project management, QA, infrastructure and the cost of someone still answering the phone in three years.
Translated into projects, the bands look roughly like this:
- Template brochure site (Squarespace or Wix, one to three pages): €500 to €1,500
- Professional SME website (custom design, proper CMS): €4,000 to €8,000
- Business website with integrations (CRM, booking, multilingual, custom post types): €6,000 to €22,000
- Ecommerce: €6,000 to €25,000 and upwards with complex catalogue or fulfilment logic
- Custom SaaS or internal platform: from around €40,000, climbing with every integration and compliance requirement
Here is the part most pricing guides skip: none of those numbers are wrong. The €1,500 quote and the €40,000 quote can both be entirely honest. They describe different products with different lifespans. The useful question is never “which is the real price”, it is “which product does this business actually need”. Plenty of Irish SMEs are better served by a €1,500 template than a €15,000 custom build, and we will tell you so.
What you are actually paying for
Break a typical custom build into its parts and the shape of the bill becomes obvious:
- Discovery, requirements and technical planning: 10% to 15%
- Design and UX: 15% to 20%
- Implementation (writing the code): 30% to 35%
- Testing, code review and QA: 15% to 20%
- Deployment, infrastructure and DevOps: around 10%
- Project management and communication: around 10%
Implementation, the one activity AI is genuinely excellent at, is roughly a third of the bill. Even a fantasy scenario where AI writes all of the code for free does not halve the price. Apply the realistic 20% to 30% efficiency gain and you have shaved perhaps 10% off the project, which is comfortably inside the margin of error created by a single scope change.
Why the AI discount hasn’t arrived
The measured data is blunter than the marketing. Against vendor claims of 50% faster delivery, teams adopting AI coding tools have seen first-year costs run about 12% higher once the full picture is counted: roughly 9% additional code review overhead, a 1.7x testing burden from increased defect density, and around double the code churn.
Perception is part of the trap. In controlled measurement, developers using AI assistance reported feeling about 20% faster while actually completing tasks around 19% slower. GitClear’s analysis of 211 million changed lines found duplicated code blocks climbing sharply as AI assistants spread while refactoring activity fell, which is precisely the pattern that converts a cheap build into an expensive maintenance bill.
Cost does not vanish, it moves. A study of open-source projects after Copilot adoption found output rose, driven mainly by less-experienced contributors, while the burden landed on senior maintainers who reviewed 6.5% more code and saw their own output drop 19%. We have written before about the verification bottleneck, and pricing is where it shows up on an invoice. Generation got cheap. Confidence did not.
Where AI genuinely does save money
We are not cynics about this. AI has changed our own economics, just not in the places buyers assume:
- Discovery and prototyping. A clickable prototype in days rather than weeks makes it dramatically cheaper to discover you were about to build the wrong thing.
- Legacy comprehension. Mapping an undocumented codebase used to be weeks of archaeology billed at full rate. It is now days.
- Mechanical migrations and test scaffolding. High-volume, repetitive, easily verified work is exactly where the tooling earns its keep.
- Documentation and technical content. Always underfunded, now affordable.
The pattern is consistent: AI reduces the cost of work that is mechanical and verifiable. It does not reduce the cost of judgement about what should be built, which is the expensive part and always was.
Fixed price, time and materials, or something better
Fixed price buys certainty, but the vendor prices that risk in, typically as 20% to 30% contingency, and every change becomes a change request with its own commercial friction. Time and materials is more flexible and cheaper when scope is genuinely understood, but the numbers are sobering: around 66.5% of organisations run 30% to 40% over budget in year one, almost exclusively on time and materials contracts.
Our view, and we will happily argue it: buy discovery as a small, separately fixed piece of work first. Come out of it with a specification precise enough to fix a build price against. You pay for certainty once, at the point in the project where certainty is cheapest to buy. It is also the structure that fits Irish grant funding cleanly, which matters if you are drawing on the schemes we covered in our guide to funding a software project in Ireland.
The number nobody puts in the quote
A €8,000 website does not cost €8,000. Ongoing maintenance for a typical Irish business site runs €800 to €3,000 per year, with light retainers covering updates, backups and content tweaks starting around €40 to €80 per month. On top of that sit hosting, domains, third-party services, payment processing, monitoring, security patching, and, increasingly, AI API costs that scale with usage rather than sitting flat.
Budget 15% to 25% of the build cost annually. A team that will not discuss year two with you is quoting for a handover, not a relationship.
Red flags in a quote
- A price with no scope document behind it
- No discovery phase, or discovery bundled invisibly into a fixed build price
- “AI-powered” listed as a feature with no explanation of what it does for the user
- Rates dramatically below market, which usually means undisclosed subcontracting or no review process
- Silence on who owns the code, the repository and the hosting accounts
- No conversation about maintenance at all
If you want the longer version of that list, our guide to choosing a development agency in 2026 covers the due diligence in detail.
How to budget properly
Three steps, in order. Decide the business outcome you are buying rather than the artefact you think delivers it. Buy discovery separately and treat its output as the real deliverable. Then budget the build, add 20% contingency, and add 20% per year of ongoing cost for the life of the system.
Do that and the quotes you receive stop looking arbitrary, because you will finally be comparing the same thing.
REPTILEHAUS builds web applications, SaaS platforms, AI integrations and Web3 systems for clients across Ireland and further afield, and we are happy to talk numbers before anyone signs anything. If you want a straight answer on what your project should cost, get in touch.
📷 Photo by Kelly Sikkema on Unsplash
