93% of web designers have used an AI tool to assist with a web design-related task at least once.1 Needless to say, AI has become indispensable for website creation. Chances are that startups will rely more heavily on artificial intelligence to create their websites—maybe the process will be faster, more efficient, and handled internally. But is this a change for the better?
In this trilogy, we explore the possibilities and downsides of using AI for website creation in 3 articles. We will delve into how tech startups can leverage AI tools, whilst ensuring the uniqueness and high performance of their websites: starting from a copywriting and wireframing perspective to a web design, and, finally, a web development stance.
In this very first part of the series, we kick off by answering the question of why AI is a double-edged sword for website copywriting and wireframing.
AI for website wireframing and copywriting: A double-edged sword
3 minutes and 19 seconds to build the wireframe for the website of a tech startup, including the sitemap and copy. In the project task in Asana, the estimated time for sitemap, wireframes, and copy was two weeks. But can AI really keep up with human creativity and analytical, problem-solving skills? Here is a critical analysis of the use of AI for wireframing and copywriting. Let's go step by step.
When AI creates above-the-fold and features sections
71% of executives believe that generative AI can enable us to create more engaging experiences for customers.2
From Relume's AI-powered site builder to Midjourney—AI tools have opened up a bunch of possibilities to speed up the content creation and web design process. Why spend days on a wireframe when AI can do it for you in a matter of seconds? Why rack your brain trying to come up with creative illustrations when DALL-E is creating an average of 34 million images each day?3
From sitemap to low-fidelity wireframe
AI-powered tools are not only speeding up workflows, but they are also conquering the creative, strategic sphere of web design. In August 2023, Relume updated its wireframe library with a generative AI element. When provided with a brief company description of the website, its AI swiftly creates an editable sitemap that can be regenerated with a simple click.
But there is more: the next step consists of creating wireframes and the copy for the whole website, solely based on the brief company description provided by the user. The result includes everything a standard wireframe needs: navbar, above-the-fold sections, headers, key features sections, footer, contact form, call to action, etc.
We don't suppose that anyone would take these wireframes into the design phase without revision. First, as we only provided two sentences about the company, AI had to come up with quite a bunch of information. Information is repeated and the style of the text lacks consistency and personality.
Next, structure. The suggested layout helps to create an impression of how big each page might be, but it doesn't consider any specific requirements. However, adjustments to the wireframe can be made easily by adding or deleting boxes, headings, and other structuring elements.
So, did this go into design?
The final low-fidelity wireframe we sent the client was completely different. With the information provided by the client, we crafted the copy, and according to the project scope and the goal of the website, we added and deleted pages, sections, and placeholders for visuals.
FYI: In the next parts of the series, you will see the wireframe with the design and, ultimately, the finished website.
Lack of creativity and branded storytelling
Thus, AI helped us to get a first idea of what a possible structure might look like and gave us inspiration for the website copy. However, the proposed solution by AI lacked personalization and uniqueness, not only for layout and composition but also for texts: a problem we go into detail further down.
With the rise of AI-powered tools and the urge to launch as fast as possible, websites might take a step back in terms of creativity and storytelling. While on the one hand animations and motion design are becoming a crucial part of each website, suggesting singularity and professionalism, wireframes created with AI might be too rigid in their structure to convert them into outstanding design.
When AI takes over the keyboard
On average, website users dedicate 5.59 seconds to reading written content on a site.4
Texts used to be created by humans for humans, following the encoding/decoding model of communication by Stuart Hall with sender and receiver. Now, AI based on Large Language Models (LLM) is becoming the sender. This goes back to 1966 when ELIZA was developed: a program that used a simple set of rules to mimic human conversation, allowing it to respond to user input in a way that seemed natural and conversational.
We innocently believed that robots and neural networks would never be able to fully understand human creativity nor follow our trains of thought—then the new AI tools entered the copywriting and conversational game.
Accumulating information and generating already spell-checked texts faster than any human being, their advantages seem clear at first sight. Also, with Natural Language Processing (NLP) technology, they are equipped with the hint of creativity we have believed to be deeply human. But, hold on: as for web design, this vision of a fully creative, all-knowing tool is the stand that people from the outside might take.
Content at speed
In day-to-day copywriting tasks, AI is useful for brainstorming ideas, doing spell and grammar checks, and creating alternatives for headlines or short paragraphs. So, there is no doubt that AI tools such as ChatGPT and Jasper save time and resources. As mentioned above, the Relume AI-powered site builder provides copy for the low-fidelity wireframes in a matter of seconds.
Although 73% of consumers trust content written by AI reliability is a big issue2. AI tools crawl information from different sources on the internet without checking its veracity, if you don't provide enough information. Hence, especially statistical data always needs to be double-checked.
Unique content for success
Also, one of AI-generated texts' major limitations is their lack of uniqueness. This is restraining its use not only because of possible plagiarism concerns but also for other reasons.
- On-page SEO: Google prefers unique website content, ranking human-made, singular texts higher and giving them more visibility.
- Coherence and cohesiveness: The copy for a whole website comprises thousands of words. Not only on the same page but across multiple pages, texts have to be coherent and cohesive, with interlinks that actually make sense. Purely AI-generated website content most probably lacks this.
- Personality: When creating unique texts that appeal to the users, personality is crucial. The copy of a website has to match the tone of voice and the personality of the company, otherwise, it might irritate users instead of catching their attention.
Quality of texts at risk
So, what does this imply for the 200,121,724 websites4 out there that are hungry for content? If companies are looking to upload content quickly without worrying about style and personality, an enormous amount of text can be created with the use of AI tools.
Keeping in mind that website users only dedicate 5.59 seconds to reading written content on a site, companies might not consider it necessary to have perfect content. Hence, the quality of texts might be at risk and users might stumble across parts of the same copy several times within a sector.
This can result in a domino effect, as search engines, since many years ago, take user behavior like interactivity and time-on-page into consideration for the ranking. If you offer dull, duplicate content, this most likely will have a negative impact on user behavior and consequently on the success of your startup website.
Thus, producing unique content is crucial for SEO, as it contributes positively to search engine ranking.
In the end, copywriting is a human-to-human interaction. By adding the human touch, it gets creative, appeals to users, and their pain points, and ideally convinces them to take a certain action. If users are confronted with irritating, incoherent content or misleading information, they might lose trust and bounce.
It is difficult, if not impossible, for a company to stand out from competitors with solely AI-generated copy.
What are the consequences for startup websites?
For startups, AI-powered content tools are a double-edged sword: on the one hand, creating website content is made easier and faster. It is easy to create a basic structure for websites, which would probably work for a handful of companies.
With the unreflective use of AI-powered website builders, websites might end up being very similar to each other. Considering the importance of a well-structured website with excellent branded storytelling, there is a risk that startup websites attract fewer users, investors, and stakeholders. Thus, today more than ever, unique websites are gaining importance.
For web designers and copywriters, it is crucial to know for which steps of the web design process they can ask AI for help. Most importantly, they have to keep in mind the brand strategy to tell a compelling brand story on the website and to present tech startups the way they deserve.
In part 2 of our series, we delve into the possibilities and drawbacks of AI in web design for startups. Stay tuned for more! In the meantime, feel free to browse our case studies to get some tech startup branding and web design inspiration.
By the way, if you are impatient to see how Vivici's website ended up after all, you can also check out our Vivici branding and website project.
2 Capgemini Research Insititute