William Tunstall-Pedoe, the 56-year-old entrepreneur who helped create Amazon’s Alexa, has shared his journey from building one of the world’s most recognizable voice assistants to launching his own AI startup. After Amazon acquired his company Evi in 2012, Tunstall-Pedoe played a pivotal role in developing Alexa, which became an instant success and household name. However, by 2016, after three and a half years at the tech giant, he made the difficult decision to leave.
Tunstall-Pedoe’s entrepreneurial journey began early. At 13, he was already experimenting with mainframe computers, which sparked a lifelong passion for pushing software boundaries. After studying computer science at the University of Cambridge and briefly teaching there, he realized his true calling was entrepreneurship rather than academia. The allure was simple: create something genuinely new in software, and it could reach a billion smartphones in six months.
In 2006, he founded True Knowledge (later renamed Evi) with a vision to revolutionize how people interact with computers. Rather than relying on keyword-based searches, he wanted to enable natural, conversational queries. After several pivots—including attempts to compete with Google and focusing on SEO—the company launched the Evi voice assistant in 2012, just a year after Apple introduced Siri. As a 30-person startup suddenly competing with the world’s most valuable company, Evi caught Amazon’s attention and was acquired later that year.
The Amazon experience was transformative. The company invested heavily in Cambridge, turning the startup into a major office. Tunstall-Pedoe split his time between Seattle and Cambridge, working on what would become one of Amazon’s “biggest and most exciting secrets.” When Alexa launched, the response exceeded all expectations—it was instantly successful.
But success brought change. By 2016, thousands of people were working on Alexa, and Tunstall-Pedoe’s role had evolved far beyond the startup-building he loved. Following Amazon’s famous memo culture, he wrote a six-page memo to himself, laying out the facts: he’d delivered everything he could, the acquisition was an unambiguous success, and the product had succeeded beyond expectations. It was time to return to the startup world.
Today, Tunstall-Pedoe runs Unlikely AI, a deeptech startup founded in 2019 that focuses on neurosymbolic AI—combining powerful machine-learning models with algorithmic precision to make AI more trustworthy and reliable. While he acknowledges the advantages of working within large organizations, he believes startups are better suited for exploring unconventional ideas. As he puts it, in a big company, one manager can kill a project, but in the startup world, you only need one investor to say yes.
Key Quotes
If you create something genuinely new in software, it can be on a billion smartphones in six months and truly change the world. That’s impact.
Tunstall-Pedoe explains his motivation for choosing entrepreneurship over academia, highlighting the unique potential of software innovation to achieve massive global scale and impact rapidly.
I imagined a world where you could have that same kind of conversation with computers, which led me to found True Knowledge in 2006.
This quote captures Tunstall-Pedoe’s original vision that eventually led to the creation of Evi and later Alexa—enabling natural, conversational interactions with computers rather than keyword-based searches.
In the memo, I laid out these facts: I’d delivered everything I could, the acquisition had been an unambiguous success, and so too had the product.
Describing the six-page memo he wrote to decide whether to leave Amazon in 2016, Tunstall-Pedoe demonstrates the clarity of thought that led him to conclude his mission at Amazon was complete and it was time to pursue new challenges.
The goal is to combine the powerful but sometimes incorrect machine-learning models with the world of algorithms, where computers are almost always right. The mission of the business is about making AI trustworthy and reliable.
Tunstall-Pedoe explains the core mission of Unlikely AI, his current startup focused on neurosymbolic AI, addressing one of the most critical challenges facing the AI industry today.
Our Take
Tunstall-Pedoe’s trajectory reveals a fascinating pattern emerging in AI: experienced builders are leaving big tech to tackle AI’s fundamental limitations. His move from Amazon to Unlikely AI isn’t just a career change—it’s a bet that the next breakthrough in AI won’t come from scaling up existing approaches, but from fundamentally rethinking how AI systems work. The focus on neurosymbolic AI is particularly prescient given current concerns about AI hallucinations and reliability. While companies like OpenAI and Google race to build larger language models, Tunstall-Pedoe is pursuing a different path: combining neural networks’ pattern recognition with symbolic AI’s logical reasoning. This could be the key to AI systems that are both powerful and trustworthy. His story also validates an important insight: innovation often requires the freedom to fail that only startups provide. At Amazon, Alexa succeeded spectacularly, but exploring truly unconventional ideas requires an environment where contrarian thinking thrives—something easier to achieve with venture backing than within a massive corporation’s hierarchy.
Why This Matters
This story offers crucial insights into the evolution of voice AI technology and the entrepreneurial decisions shaping the AI industry. Tunstall-Pedoe’s journey from creating Alexa to founding a neurosymbolic AI startup highlights a critical trend: experienced AI pioneers are increasingly leaving big tech to pursue more ambitious, unconventional approaches to artificial intelligence.
His new venture, Unlikely AI, addresses one of the industry’s most pressing challenges—AI reliability and trustworthiness. As machine learning models become more powerful but occasionally produce incorrect or “hallucinated” results, the combination of neural networks with symbolic reasoning (neurosymbolic AI) represents a promising path forward. This approach could be essential for deploying AI in high-stakes applications where accuracy is paramount.
The story also illuminates the trade-offs between innovation at scale versus startup agility. While Amazon provided Alexa with instant exposure and massive resources, Tunstall-Pedoe’s experience suggests that truly novel AI breakthroughs may require the freedom and risk-taking culture that only startups can provide. For the AI industry, this raises important questions about where the next generation of transformative AI technologies will emerge—from tech giants or from nimble startups led by experienced innovators.
Related Stories
- Google Workspace’s Robert Frati Left for AI Startup Gemini Amid Tensions Over AI Strategy
- Sam Altman and Jony Ive Partner on AI Device, Target $1B Funding
- The Future of Work in an AI World
- Microsoft AI CEO’s Career Advice for Young People in the AI Era
Source: https://www.businessinsider.com/alexa-co-creator-why-i-quit-amazon-launch-ai-startup-2026-1