Neural Notes: Stop building AI startups with “the same crap” as everyone else

Welcome back to Neural Notes, a weekly column where I look at some of the most interesting AI news of the week. In this edition: warnings for startups relying too heavily on generic AI models and how AI has changed the relationship between VCs and founders.

The proliferation of startups riding the gen AI wave over the past two years has been industry-changing. But while the technology is undeniably transformative, for both businesses and VCs, it’s a double-edged sword.

Are too many founders leaning on off-the-shelf AI models without building anything genuinely innovative?

This was the critique from Pedram Mokrian, a seasoned venture capitalist and lecturer at Stanford University, during Assembly 2024: A VC Catalyst Conference in Melbourne this week.

Organised by LaunchVic and The Wade Institute, the conference brought together investors and founders to explore the future of venture capital and the Australian startup ecosystem.

During a panel titled Startup Investment 2025+, Mokrian pointed to a growing trend of ‘AI’ companies that fail to differentiate themselves in an increasingly competitive market.

“One thing I’m worried about with the advent of AI is most of the startups will look more similar than dissimilar because the tools are now available to everyone in the same ways,” Mokrian said.

“We all use the same generative AI for assets, for images, for code, and things will look more similar than dissimilar for a while as we navigate this next phase.”

Mokrian warned that startups relying entirely on generic AI models are failing to address what he called the “network effect problem” – the ability to create scalable, unique value through defensible advantages like proprietary data or specialised applications.

“Really large outcomes, venture-backed outcomes, require some amount of defensible network effects, which is unclear how using the same crap everyone else uses gives you defensible network effects,” Mokrian said.

AI’s effect on the relationship between founders and VCs

The rise of AI has also contributed to the change in traditional dynamics between founders and VCs. Thanks to advancements in cloud computing, open-source AI tools, and large language models (LLMs), the cost of launching a startup has plummeted.

Founders now have access to powerful tools without needing significant upfront capital. This, combined with alternative funding sources, has somewhat shifted the balance of power.

“Founders need VCs less than ever before,” said Aaron Batalion, co-founder of LivingSocial and now an active angel investor.

Panelists agreed for VCs to remain relevant, they need to offer strategic value beyond financing.

“If you don’t need my money, what else?” Batalion said.

He went on to cite emerging approaches where VCs offer unique resources like GPU access to attract founders who might no longer rely on traditional funding routes.

“There are some funds in Silicon Valley that just give people access to GPUs… If that’s their special sauce, let us invest in your company,” Batalion said.

AI startups solving real problems

Despite these critiques, Assembly 2024 did highlight examples of startups using AI to solve deeply entrenched problems.

As part of the keynote panel, hosted by yours truly, we heard from the founders of Andromeda Robotics and InTruth AI in a panel titled Empathy, Technology, Connection and You.

Andromeda Robotics, founded in 2022, has developed Abi, a humanoid companion robot designed to provide support and companionship for older adults in aged care facilities and children in hospitals.

Equipped with AI and machine learning, Abi can recognise faces, express emotions, and assist with daily tasks. Abi can be equipped with different personalities to suit individual organisations – anything from sassy to Christian – to combat loneliness and improve quality of life.

InTruth AI uses consumer-grade wearables and proprietary AI to track emotional states in real-time, offering users actionable insights to better understand and manage their emotions.

By translating raw data from wearables into emotional patterns, InTruth has the potential to revolutionise mental health support across individual and organisational contexts.

Throughout the day we also heard about other AI-powered startups making tangible differences.

One standout case was an AI-powered pest management platform – RapidAIM – that uses predictive algorithms to help farmers identify and mitigate infestations before they occur.

Another example was an AI-driven computer vision platform for meat processing facilities, which enhances safety protocols and operational efficiencies.

Unlike generic applications of AI, these solutions address urgent, niche problems. They’re evidence of how the thoughtful deployment of technology can lead to real transformative change in the world.

The key takeaway from the day, at least in regards to AI, was that founders who merely layer AI on existing solutions – or rely on other company’s out-of-the-box solutions – may struggle to stand out.

However, startups that take the time to innovate, leveraging AI to address specific, high-impact challenges, have the potential to transform industries and create real value.