Twelve ambitious AI startups have been selected to compete in the AI Pitch Competition. The finals are on November 7, 2024, and Innovation Origins will portray each contestant in the run-up to that event. The AI Pitch Competition is a Brabant-based organization that aims to spotlight the most innovative AI solutions, offering startups the opportunity to present their ideas, connect with industry leaders, and accelerate their growth. Today, we show what Touchpulse has in store for the world. Liam Geschwindt, co-founder, answers our questions.
Touchpulse‘s mission is to empower individuals with visual impairments to live a life unhindered by their challenges. By harnessing the power of AI-driven solutions, Touchpulse crafts a tool that captures and communicates the critical details of one’s environment. Touchpulse is developing a product and deep learning model in-house in collaboration with the community.
1. What specific AI technology or algorithm is at the core of your solution, and how does it differentiate from existing approaches in the industry?
“Our solution relies on the ability to ask questions to your navigation system. This feature goes far beyond just voice control like “navigate to X”, and, depending on how our users will use the app, will allow you to ask questions about your surroundings, how to cross streets, etc. This system builds upon novel language model techniques such as those developed by Google (Gemini) and OpenAI (ChatGPT).
This unlocks a range of new autonomous navigation options for our target audience of people with visual impairments. After some time, we will optimize models to answer more accurately the questions our users may have and improve automatic guidance systems.”
2. How scalable is your AI solution, and what are the primary challenges you anticipate in bringing it to a global or large-scale market?
“The limit of our scalability is either that of large providers of LLM technology (such as Gemini and OpenAI) since our users will regularly rely on the services of these providers, or that of our own compute infrastructure, which we can add-on if limitations in the infrastructure of our cloud providers arise.
Our main challenge lies in raising awareness among the global group of visually impaired people, our VIPs. Since their online presence is limited, we spend more time and effort reaching this audience than we would for general navigation users. Beyond this, cultural differences in how people interpret and expect functions in navigation solutions make it more difficult to align with everyone’s specific needs. Our solution will be able to – unlike traditional navigation systems – provide tailored information per person by adapting over time.”
3. How does your startup address potential ethical concerns related to bias, fairness, or transparency in AI decision-making? Can you provide an example of how you mitigate such risks?
“While striving to advance social, economic, and infrastructural equity, our navigation system also grapples with key ethical challenges. Addressing these challenges requires a well-considered approach, particularly as we balance progress on some Sustainable Development Goals (SDGs) with unintended consequences on others.
Our primary mission to enhance accessibility for people with disabilities (mainly visually impaired persons, or VIPs) positively impacts SDGs 10 (reducing inequalities) and 11 (making cities inclusive). However, this technological solution generates energy consumption that negatively affects SDG 3.9, which addresses pollution. The ethical challenge here is minimizing our carbon footprint while maintaining the inclusivity and accessibility we seek to enhance.”
4. In what ways do you believe your AI solution can positively impact society? Are there any unintended societal risks or challenges you foresee, and do you prepare to handle them?
“Our product fosters greater economic participation and social inclusion for VIPs, aligning with SDGs 10.2, 10.3, and 11.2. However, an ethical dilemma arises in ensuring that our technological solution does not exacerbate digital inequality or create a dependency on high-tech systems that may be out of reach for economically disadvantaged populations.
To tackle this, we aim to make our product as affordable and accessible as possible, ensuring that users across different economic backgrounds can benefit. Additionally, we focus on usability and inclusivity in design, ensuring that the product is intuitive and not overly reliant on expensive or proprietary technologies.”
“Every week there are 10 new speech-to-text, text-to-speech, LLM / NLP models that are ‘better’ than the last. Figuring out our winners and building upon them is difficult in such a volatile market.”
Liam Geschwindt
5. What has been the biggest hurdle in launching your AI startup, and how did you overcome it? How has the current AI landscape shaped your business model and growth strategy?
“Our biggest challenge is understanding how to align our system with the needs of our audience. VIPs have a different set of needs compared to our own, and building empathy and knowledge about their thought processes becomes a challenge because of this. We spend much time with our early adopters to continue smoothing out the integration with their daily travel.
Since our AI will take on this role later, we must have a strong understanding, so we can ‘teach’ it to our system. We couldn’t scale if we had to personally guide our users. Another challenge is deciding and building upon the systems that are being released. Every week there are 10 new speech-to-text, text-to-speech, LLM / NLP models that are ‘better’ than the last. Figuring out our winners and building upon them is difficult in such a volatile market. We welcome this, though, since progress is steady and each step improves our users’ experience.”
6. How do you prepare for the increasing regulatory frameworks around AI, such as GDPR, AI Act, or other data privacy laws? How does this impact your innovation and development process?
“We have taken extra security measures such as separating user information and usage data, user access authentication, administrator-only access to usage information, etc. Furthermore, we are researching, together with other student startups at TU/e, which AI and privacy laws apply, and what they concretely mean, so we can adapt and protect our users’ privacy. This does not negatively impact our innovation process. The only cost is some time, but it does not impede our development.”
7. How do you see your technology evolving over the next 5 to 10 years, and what role do you expect your company to play in shaping the future of AI?
“The future of AI is bright, but we see some space where AI is still lacking: Personalization. By personalization, we mean adapting to a user by interacting regularly and over a longer period of time, learning their mannerisms and preferences, and trying to adapt to that automatically. Our system will do exactly that. Our team is hence laser-focused on building the infrastructure to learn from users rapidly (perhaps in real-time), and we see this unlocking new opportunities in the future. Think of personal assistants, integrating more parts of your digital presence so the system knows where you want to go beforehand, and expanding to the general public, so our users can say ‘We use the same tools everyone else does’.
We have a strong AI team in-house that will work on these challenges over the coming years. If we are still around in five years, which we work hard for, I can see us being pioneers in this field.”
8. Why are you going to win the AI Pitch Competition?
“Our solution uses all the new AI technologies: LLMs, speech recognition and synthesis, Retrieval Augmented Generation, tool-calling, and more. We will not only copy existing work but also build next-generation AI tooling while making a real-world positive impact on disadvantaged groups of people. We are now one of twelve finalists, and with our positive impact on SDGs and a clear solution to a clear problem of a clear target audience, we will win the AI pitch competition. Not only because we use AI innovatively but also because we use it in an effective way that can immediately improve the lives of millions around the world.”
Collaboration
Related Posts
