Nest founder Tony Fadell on next generation of promising deep tech startups | TC Disrupt 2024

The iPod, Nest, and Early Career Influences

• The conversation starts with a casual discussion about the speaker's beard and appearance, with a mention of Mike Butcher and a joke about a "COVID beard" (12s).
• The speaker is asked if they feel annoyed talking about the iPod, similar to musicians who get tired of playing the same old songs, but they don't feel that way (1m0s).
• The speaker mentions that the 23rd anniversary of the iPod launch and the 13th anniversary of the first Nest product launch both occurred recently, which is a coincidence (1m34s).
• The speaker's father was a salesman who sold jewelry and watch parts, and they learned the importance of sales and marketing from him (2m8s).
• The speaker notes that they learned to develop sales and marketing skills while working at General Magic and Phillips, as they realized that building a product is not enough if you can't sell and market it (2m22s).
• The speaker's father worked for Levi's jeans for 34 years, and they mention that while Levi's jeans may sell themselves, they still need to be marketed and sold (2m49s).
• The speaker recalls their first product in college, where they spent all their money on development and had none left for sales and marketing, teaching them the importance of allocating sufficient funds for marketing (3m8s).
• The speaker notes that the math has changed with the rise of social media, but they still needed 5-10 times more money for sales and marketing than for product development (3m22s).

General Magic and the Importance of Timing

• General Magic was a company that faced numerous issues, including branding problems, and was ahead of its time in creating technology similar to the iPhone 15 years before its release, but ultimately failed due to a lack of understanding of the problems it was trying to solve and poor timing (3m34s).
• The company was working on solving problems such as email, e-ticketing, downloadable games, animations, and emojis, but the technology and market were not yet ready (4m16s).
• The importance of hitting the market at the right time and understanding the problem being solved is crucial for a company's success, as demonstrated by General Magic's failure despite having seminal ideas (4m51s).
• The team at General Magic included people who later created Android, and the company's work laid the foundation for future technologies (4m57s).

Working with Larger Companies and Understanding Growth

• Working with a larger company can be beneficial for entrepreneurs, as it provides experience and a model for how to operate, which is essential for growing a startup (5m36s).
• The way a company operates changes significantly as it grows, with different cultures and communication styles emerging at different stages, such as 5-10 people, 15-20 people, 30-40 people, and 120 people (6m35s).
• Understanding these "break points" is crucial for entrepreneurs, as they are fundamental human nature issues that can make or break a company's success (6m49s).
• Over 200 startups around the world have received direct investment, and it's essential to teach these entrepreneurs about break points, organization structure, communication, and human nature issues that come with growth and building things (6m57s).

Learning to Delegate and Trust as a Leader

• As an entrepreneur and leader, it's crucial to understand the importance of letting go of things and trusting the people on your team, which can be a difficult lesson to learn (7m45s).
• Tony Fadell had to learn this lesson when he became a CTO and vice president at Phillips at the age of 23 or 24, and he had to build a team of hundreds of people despite having no prior management experience (7m57s).
• Fadell failed as a manager for the first year and had to get training to learn how to delegate and trust, and he eventually became more direct and knew where to push and what not to push on as he gained more experience in the industry (8m57s).
• Fadell's approach to management is not about being easier, but about being more direct and knowing exactly what to say at certain times, and he believes in giving people enough rope to learn and grow but not enough to fail (9m22s).
• Fadell's experience has taught him that growing teams and leaders requires letting them go and learn from their mistakes, and he now focuses on the things that really matter and provides mentorship from time to time (9m50s).

Detail-Oriented Leadership and Mission-Driven Focus

• There are two types of people who are detail-oriented: those who are egocentric and try to control others, and those who are mission-driven and focus on critiquing work to achieve a goal (10m37s).
• Mission-driven individuals are not being overly critical, but rather keen on details to ensure the delivery of high-quality products that meet customer needs (11m21s).
• Leaders who are detail-oriented and care about the mission can push their teams to achieve great things, but it's essential to know which details matter and which don't (12m2s).

The Nest Acquisition and Culture Clash

• Working at a startup and a corporation have distinct differences, and when a startup is acquired, there can be a breakpoint where the culture changes overnight (12m42s).
• The acquisition of Nest by Google led to a culture change, which was partly due to mistakes made by the management team, resulting in a loss of the original culture (12m57s).
• Most mergers and acquisitions (M&As) fail due to significant cultural differences between the companies involved, with 75-85% of M&As failing for this reason (13m7s).
• The transition from "founder mode" to a more established company can be challenging, and if not managed correctly, can lead to a loss of motivation and focus among team members (13m23s).
• In hindsight, the Nest deal can be seen as part of the 80% of M&As that fail, as the company's culture and mission were not preserved after the acquisition (13m53s).
• A culture clash occurred when a company was acquired, as the acquiring company did not honor the prenuptial agreement and instead imposed their own culture, leading to a dramatic change in the work environment (13m58s).
• The culture at Apple and Google was dramatically different, with Apple being a place where everyone was critical to the programs and you couldn't hide, whereas Google had a more relaxed culture with 20% time, allowing employees to work on side projects (14m29s).
• At Apple, employees were expected to be critical to the programs and couldn't hide, whereas at Google, employees could hide and were not always held accountable for their work (14m53s).

Google's Culture and the Spread of Entitlement

• The culture at Google allowed employees to have a lot of freedom, with some employees taking advantage of this by not showing up to work or only showing up for lunch, which led to a sense of entitlement (15m5s).
• This culture of entitlement has spread to the rest of Silicon Valley, with many companies adopting a similar culture, which is a major turn-off for some people (16m30s).

Moonshots, Deep Tech, and Investment Challenges

• The speaker believes that moonshots, or ambitious projects, cannot happen in the context of a large corporation like Google, as they are often given unlimited resources and are not constrained, leading to a lack of accountability and a focus on pet projects rather than profitable businesses (17m5s).
• The speaker cites the example of Waymo, a self-driving car project that has spent tens of billions of dollars but is still not a profitable business, as an example of how moonshots can fail to deliver in a large corporation (17m26s).
• Many startups die along the way, and even successful companies like Google and Alphabet have had projects that didn't pan out, highlighting the challenges of deep tech investments (17m43s).
• Deep tech companies require a long runway, a constrained budget, and a clear understanding of their goals, with near-term milestones and a focus on making progress rather than trying to achieve overly ambitious goals (18m2s).
• Teams working on long-term projects can become disengaged and demotivated if they don't see progress or receive customer feedback, making it essential to have a clear plan and achievable milestones (18m40s).
• The failure rate for M&A deals is high, and it's likely even higher for deep tech investments, with 80-85% of VC portfolios likely to result in a loss or a write-off (19m25s).
• However, the successes in deep tech can be significant, with fundamental and disruptive technologies having a major impact, making the investment worth the risk (19m41s).
• Deep tech teams often have constrained resources, but they are focused on solving a specific problem and are working hard to achieve their goals, with some teams achieving success despite the challenges (20m8s).
• The calculus for investing in deep tech is complex, with a long time horizon and significant risks, but understanding the fundamental technology and first principles can help investors make informed decisions (20m47s).

Managing Capital and the Success of Deep Tech Companies

• Properly managing capital, both human and financial, is crucial for success in deep tech investments, allowing investors to navigate the challenges and achieve their goals (21m0s).
• Capital can be raised by achieving small milestones, which attracts other investors and people to invest, as seen in companies like Diamond Foundry (21m2s).
• Companies like Oranus, which started in 2015, are working on AI-driven cancer vaccine and cancer drugs, and have already completed Phase 1 trials (21m24s).
• Other companies, such as Plumerai and Tiny AI, have been working on AI-driven projects since 2017, and are now making a big impact (21m42s).
• These companies are using AI in deep tech spaces, such as drug development, and are making a profit or making a significant difference (22m1s).

The LLM Revolution and the Future of AI

• It took a long time, 5-10 years, for these companies to achieve their goals, and it's expected that in the future, the large language model (LLM) revolution will be seen as a bubble to some extent (22m15s).
• LLMs are great for certain things, but they are not the only solution, and other types of models and combinations of models are being used behind the scenes (22m25s).
• Artificial specific intelligence (ASI) has been working well, as seen in the first AI thermostat developed by Nest in 2011, and ASI does not hallucinate like LLMs (22m51s).
• The timing of AI development is crucial, and companies need to be talking about AI to be relevant, but they also need to be aware of the limitations of LLMs (23m8s).
• LLMs are trying to be a general solution, but they are not always the best choice, and businesses need to focus on what works today, rather than trying to make science fiction happen (23m23s).

Transparency and the Need for Regulation in AI

• There is a need for a LinkedIn-like platform for AI agents, which would provide information on what they were trained on, their limitations, and their potential biases (23m57s).
• This platform would allow businesses to make informed decisions when hiring AI agents, and to know what they are capable of and what they are not (24m20s).
• There is a need for transparency in AI systems, similar to LinkedIn, to understand what they are and how they work, and governments should have regulations to ensure transparency on data, errors, and other aspects to build trust in using these systems in business and customer service (24m32s).
• The lack of transparency in AI systems can lead to problems, such as the 90% of doctors who use ChatGPT to create patient reports experiencing hallucinations, which can have serious consequences (25m5s).
• There is a need to understand what AI systems are being used and what problems they can cause, and having transparency on data and hallucinations is crucial to avoid disasters (25m37s).
• The "Black Box problem" in AI refers to the lack of understanding of how AI systems work, and it's a significant issue that needs to be addressed (26m2s).

Alternative Approaches to AI and the Limitations of LLMs

• Some companies, like Plume, are working on smaller AI models that are more accurate and efficient, such as an 8-megabyte model that can do video analysis and person recognition (26m34s).
• Apple's approach to AI is working because it focuses on building smaller, more accurate models that can be improved upon, rather than trying to create a massive model that does everything but doesn't work well (27m8s).
• There are fundamental problems with the current approach to AI, and it's unlikely to improve significantly, which is why some people are exploring new areas like Quantum AI and Quantum LLMs (27m41s).
• Quantum AI and Quantum LLMs are being explored as a potential solution to the limitations of current AI systems, but it's a complex and separate topic (27m57s).

The Realities of VR and AR Technologies

• The current state of Virtual Reality (VR) technology is still plagued by human factors problems, despite advancements in fidelity, making it unsuitable for everyday use (28m40s).
• VR has great potential for certain episodic uses, such as training doctors or collaborative 3D design, but it's not meant for prolonged wear (28m45s).
• A company called Gravity Sketch, which allows for collaborative 3D design in a virtual world, is a promising example of VR's potential in specific applications (28m59s).
• Augmented Reality (AR) is considered a more viable technology, but VR's limitations make it impractical for widespread adoption (29m21s).
• Despite four generations of VR development since 1988, the technology still faces the same human factors problems, including bulkiness and social awkwardness (29m32s).

Humanoid Robots and Their Potential Applications

• Humanoid robots, like those developed by companies such as Agility, have potential for specific tasks, but their development and training require a vast amount of data and time (29m59s).
• The use of humanoid robots in highly constrained cases, such as logistics, is more feasible, but their application in household chores will take years to develop and become cost-effective (30m21s).