Coding becomes cheaper. But your project probably won't.
The decoding of the first human genome cost three billion dollars and twelve years. Today, sequencing costs less than 200 dollars and takes just a few hours. However, this drop in price has not made genome research cheaper - it has made it more ambitious. Questions that were previously simply not feasible are now standard tasks. The savings potential of a technology rarely disappears. In most cases, it simply moves into a new scope.
It is precisely this mechanism that we are currently experiencing in the development of digital platforms - and it is changing project costing in a way that most decision-makers have not yet factored in.
The calculation that doesn't add up
According to a recent analysis by Andreessen Horowitz , software development is by far the most dominant AI use case in the enterprise sector, with a clear lead over all other application areas. For many, the obvious conclusion is that digital projects are now becoming more favourable. However, the genome example tells a different story.
We have tested this thesis on a current project. We are currently developing a comprehensive digital platform for the marketing and digital sales support of meetings, incentives, congresses and events for a tourism region. The total budget is in the low six-figure range, of which the technical implementation accounts for around 40%. However, it would be too short-sighted to only look for potential savings there. We therefore also analysed concept & UX/UI (around 20 %), content (around 20 %) and project management (just under 20 %).
Here are the learnings:
1. development: Writing code becomes cheaper, but software development only to a limited extent.
Of course, AI-supported coding is already being used at MASSIVE ART. In some cases, this radically simplifies the way our developers work. But: the devs cannot be replaced in the foreseeable future. A lot of energy still goes into quality assurance and solving individual problems outside of standardised routines. And someone has to write the prompts. So the shift is moving from manual development to product management.
2. concept & research: AI has not accelerated the concept phase, but deepened it.
User research, which previously had to remain superficial for time and budget reasons, could be elevated to a new quality dimension with AI support. Same amount of time, significantly better database.
3. UX & design: Modern design tools with integrated AI functions enable noticeably more output with comparable effort.
Design systems can be documented more consistently and variants can be evaluated more quickly. What remains unchanged is the strategic judgement that distinguishes a design from a truly effective concept. This is and remains human expertise.
4. project management: This is where AI has been least noticeable so far, and this is likely to remain the case.
AI-supported documentation in modern collaboration tools promises more transparency. But leadership, decision-making authority and stakeholder management remain what they have always been: People business. This cannot be automated away.
5. content: AI-generated texts and media offer clear savings potential, but caution is advised.
AI is already a game changer, especially for translations and structured content such as SEO texts. But who develops the content concept and writes the prompts? What's more, by publishing generated content without checking it, you risk interchangeability - and that is the most expensive thing in a competitive market. So here too, human expertise cannot be rationalised away.
What really happens
The savings potential of AI is real. However, it does not disappear from the project - it migrates.
In terms of technical implementation, the increase in efficiency leads to more complex backend architectures, cleaner interfaces and more code security through automated testing. In terms of content, it means more language versions, deeper personalisation and an editorial workflow that was simply not affordable in the past.
The volume remains. The scope is growing.
There is also a factor that is almost completely absent from the public debate: AI also generates new work. Prompt engineering is expertise. Output evaluation is expertise. Deciding when an AI-generated result is good enough and when it needs to be revised is the actual task. And this cannot be automated.
What this means for your next investment decision
Strategic decision-makers must not allow themselves to be blinded by the promise of supposedly more favourable digital projects. Instead, focus on what is suddenly possible for the same budget, but was previously beyond the realms of possibility. Investing in clean, modular structures today (preferably based on open, licence-free technologies without dependence on individual providers) will multiply the AI leverage of tomorrow.
The real question is not a few percentage points less in your budget - it's your strategic priorities.
A conversation worth having
Discuss with us which potentials in your digital projects are suddenly within reach thanks to cost savings through AI - concretely, without obligation and based on your actual digital strategy.
What we can do for you
Let us evaluate your data infrastructure before you start your next AI experiment. In a non-binding discussion, we will check your data readiness and outline the architectural path to scalable AI value creation.
We look forward to the dialogue with you.
