About
Activity
11K followers
-
Peter Thum shared thisWhat makes a voice AI feel like it actually understands you? Kim Isenberg and I spent an hour with Andrew Ettinger and Olya Ossipova from Hume AI to look into it. The conversation covered how they test voice models, what gets lost when a conversation is reduced to a transcript, and the harder question of how any AI company can honestly evaluate a market it also competes in. If you build voice products, buy them for your company, or just want to understand where the frontier of emotional intelligence in AI is, it is worth your time. Link to the full interview in the first comment.Peter Thum shared thisHow do you tell whether a voice AI actually understands how you feel? Peter and I sat down with Hume AI’s Andrew Ettinger and Olya Ossipova to talk about what gets lost when a conversation is reduced to a transcript: hesitation, frustration, urgency, and tone. We explore how Hume AI tests voice models, where automated evaluation falls short, and why human judgment is still central to measuring a good conversation. We also ask how an AI company can evaluate an industry it competes in. Watch our full conversation below (link in comments):
-
Peter Thum shared thispretty poignant.Peter Thum shared thisThe race for AGI, told like a movie trailer. Script by Sherpa/Pocket FM. Video by Seedance 2.5. by the way, here are Startup Resources you'll love: • 10k Investor List: https://lnkd.in/eQHXjjpP • 20+ Financial Model Templates: https://lnkd.in/eniYg7JH • Claude workflows, prompts and templates to make it your coworker: https://lnkd.in/ettUa_YU • 45+ AI Decks (Eleven Labs, Artisan, etc.): https://lnkd.in/etqM3-ik • OpenAI’s New Playbook for Product-Market Fit in AI Startups https://lnkd.in/dy-HgVtX • 2,500+ Angel Investors Backing AI Startups - https://lnkd.in/djjN-TD2 • 50 AI Agent Startup Ideas for 2025 - https://lnkd.in/dy9E-MCn • Family Offices list: https://lnkd.in/e72hVhph • How to use Lovable instead of PPT presentations - https://lnkd.in/eKfM7VRa 50 more here: https://lnkd.in/d_NvAM_9
-
Peter Thum shared thisReally interesting conversation today with (Mick) Michael Costigan, VP of Salesforce Futures, and my cofounder Kim Isenberg at Dreamforce 2026. We covered a lot of ground on how the future of AI actually plays out. Scenarios, not predictions. What has to be true for each to unfold, and what Salesforce and its clients are doing to prepare for more than one. We'll share the interview soon.
-
Peter Thum shared thisLinkedin peeps. This was such a fun conversation. Great guy. Cool stuff -- our chat w/ Yaron Inger cofounder of LTX and my cohost Kim Isenberg. Are you into AI dev? Watch! Do you like to learn? Watch! Want to be a part of the future? Watch! Want to hear people have fun talking about cutting edge tech? Of course you do, right. Links in comments.Peter Thum shared thisWhen does an AI video generator become a real world model? That question was at the center of our conversation with Yaron Inger, Co-Founder and CTO of LTX. LTX describes its latest open-weights model, LTX-2.5, as a world model. That is a major claim. Generating a convincing video is one thing. Representing how objects behave, persist and interact over time is a much harder problem. In this clip, we discuss what a model must be capable of before the term “world model” becomes meaningful. Does it understand why a ball falls from a table, or does it simply generate frames that look plausible? Can its simulated dynamics provide useful training data for robots? How does LTX test physical consistency, object permanence and causality? These questions matter far beyond AI filmmaking. Video models could eventually become powerful simulation engines for robotics, autonomous systems and physical AI. But that future depends on whether they learn more than visual patterns. Peter Thum and I explored this distinction with the person responsible for much of the technology behind LTX-Video and LTX-2. Watch the clip below and find the full conversation here:
-
Peter Thum shared thisThe sovereignty question (who controls your data and infrastructure) gets harder the moment agents start acting on your behalf. Data residency is the easy part — controlling what an autonomous system can do, and proving it, is where most frameworks fall short. Kim Isenberg’s conversation with Google Cloud’s Wieland Holfelder gets into exactly that issue.Peter Thum shared thisEurope talks a lot about AI sovereignty. But the rise of autonomous AI agents makes the question much more urgent: Who controls the data, infrastructure and permissions when AI systems begin acting on our behalf? I recently sat down with Dr. Wieland Holfelder, VP Engineering at Google Cloud, to discuss what sovereign AI could actually look like in Europe. We talked about the security risks created by autonomous agents, Google Cloud’s cooperation with Germany’s Federal Office for Information Security, and whether Europe can maintain control over critical AI infrastructure without falling behind technologically. The full interview will be released soon. Here is a first preview.
-
Peter Thum shared thisThis is one of the most concrete and interesting conversations we've had on what enterprise AI economics actually look like at scale. Kim Isenberg and I sat down with Jonathan Cornelissen and Yusuf Saber at DataCamp to dig into this. 10M+ hours of AI-mediated learning. Tens of millions in token costs. A 27B open model beating frontier on the evals that matter. Enterprise clients who need to swap the underlying model in minutes. If you're building AI into a real product with real users, watch this. Full interview in the comments.Peter Thum shared thisExplaining how the AI tutor actually works with Yusuf Saber in interview by Kim Isenberg and Peter Thum from Superintelligence. Link to full interview in the comments. There were 10M+ hours of learning on DataCamp last year, and our goal is to shift all of it to the AI Tutor. This creates massive opportunities and challenges. Top 10 things we cover: 1. The importance of open weight models in an educational context: open vs frontier, not in theory but in the context of helping millions of learners. 2. How we're thinking about (tens of) millions of dollars in AI tutor token costs, and why cost has become a key engineering problem. 3. Why a 27B open model is beating frontier models on our teaching evals, and why we still can't put it in production. 4. The one bottleneck stopping us from switching to open weights tomorrow. 5. What it takes to teach an AI to teach: 100+ tutor behaviors if you used them all at once. 6. How the tutor decides, at every learner response, which teaching behaviors matter right now, and which to leave out. 7. How we can swap the underlying model for millions of learners in minutes, and why enterprise clients with approved-model lists make that a core feature, not a nice-to-have. 8. Evals: 60% of our quality bar was sourced by a system (we call it Sentinel) that watches for learner friction. 9. What 300,000+ learners on the AI Tutor taught us: higher conversion, higher satisfaction, and an engagement distribution that shifted in a way we didn't expect. 10. Why we're not building an AI tutor for just data/AI skills. We're building the best AI teacher, full stop, and data/AI skills are just where the gap is biggest today. We have several principal and staff AI engineering positions open. Get in touch if you're passionate about building the #1 AI tutor.
-
Peter Thum shared thisEveryone chases GPUs. The real bottleneck is usually the data layer. I learned the hard way when we were building Muse. Our lead engineer set up a versioning system for images — storing the smallest file size that maintained quality and only updating when needed — because fetching was what killed page rendering speed. The compute wasn't the problem. Moving and retrieving data efficiently was. Our Editor in Chief of Superintelligence AI LLC, Kim sat down with Sven Breuner from VAST Data to get into exactly this problem at scale. If you're an early-stage founder building anything with significant media or model infrastructure, this isn't just an enterprise conversation. The architecture decisions you make now about your data layer will either accelerate you or tax you for years. Worth your time. Link in comments.Peter Thum shared thisGPUs are expensive. But what happens when they’re waiting for data? I recently sat down with Sven Breuner , Field CTO International at VAST Data , to discuss one of the most overlooked bottlenecks in AI: the data layer. Because building better AI infrastructure isn’t only about adding more compute. It’s also about keeping that compute fed with data. Here’s a short clip from our conversation. Full interview in comments:
-
Peter Thum shared thisMy cofounder Kim went to Munich last month for XPENG's Global Brand Day and sat down with Dr. Xianming Liu, Head of XPENG's General Intelligence Center, for an hour on where physical AI is actually going. Worth your time if you care about any of the following topics: 1. how VLA 2.0 combines vision, language, and vehicle actions in a way most Western coverage misunderstands; 2. when driverless robotaxis actually reach Europe; 3. whether the same AI foundation can drive a car, walk a humanoid, and fly a car (XPENG builds all three); 4. how China thinks about deploying physical AI at scale; and, 5. what the West keeps getting wrong about Chinese AI. Liu came out of Meta and Cruise before joining XPENG. He answered every question directly. Kim doesn't do softball interviews. Link in the comments.Peter Thum shared thisI sat down with Xianming Liu , Head of XPENG Group’s General Intelligence Center, for an exclusive interview about what VLA 2.0 can actually do, where autonomous driving still fails, when this technology could reach Europe, and how far XPENG is from putting cars without anyone in the driver’s seat on public roads. YouTube-link down below:
-
Peter Thum shared thisToday the Musizi University Foundation fundraiser hit $660K. Waaaaayyyyy past the $500K goal and the university will soon welcome its first entering class of students. Back in January of this year, my friend and fellow Claremont McKenna College alum Douglas Peterson — a trustee of our alma mater — co-hosted a fireside chat with me in Manhattan about a university that didn't yet exist. The room was packed with people who'd never heard of Musizi before that night. At that point, the fundraising campaign had raised just about a third of their target. Every institution you know by name began this way. A small group of people believed in an idea when there was no proof, acted without permission, and lent a hand when no one could promise it would work. Musizi began this way. The founding team of Neil Collins, Takako Mino, PhD, Elaine Alowo-Matovu and Sheila Nabanja, Ph.D didn't wait. They invited, they showed up, they told the story and kept selling the vision. Michael Kleinheinz has since joined to help push it forward. A huge thank you to everyone who gave, introduced, or simply showed up that night and in the months since. Thank you to Claremont McKenna College for helping us with outreach. Thank you to the team at Civic Hall and The Fedcap Group for giving our event an amazing venue, delicious food and hot cider on a freezing cold NY night. Collectively, you proved that a university on the other side of the world with no track record and no guarantees could still earn belief from a room of strangers in Manhattan and then from many rooms of people in cities all across America. I must say with no small amount of pride that my alma-mater figured out ways to help us that might be considered "no no's" at other institutions. CMC folks found a way to say yes, found a way to support something deserving, hopeful, special and new. After all, CMC was in just the same tenuous spot a mere 80 years ago. So, with the vision of an alumna (Takako), who had worked and risked and given her all, the Claremont McKenna College Development Office and the Alumni Association were the catalyst of this effort and reminded me why I'm proud to say that I'm a Stag, that it's where I got a degree and played on their gridiron for 4 amazing years. Congratulations to the whole Musizi University team. This is what conviction, hard work and entrepreneurship looks like. Link to the University website and their foundation site (where you can donate) in first comment below. Check it out!
-
Peter Thum liked thisPeter Thum liked thisStanford used AI to erase a real male student from an ad and replaced him with a fictional black woman. The student says he feels, “silenced and erased.” Stanford University has admitted to violating its own AI policy after using generative technology to erase a Latino student from an advertisement and replace him with a fictional Black woman. The controversy erupted when Billy Ramirez, a Stanford senior, discovered that a 2024 dining hall photo of himself and two classmates had been heavily doctored for official campus banners. In the newly generated promotional image, Ramirez’s face, race, and gender were completely replaced by an AI-generated Black woman. Furthermore, the AI was used to significantly slim down the faces and bodies of his two remaining classmates. Ramirez expressed deep frustration over the digital manipulation, stating that seeing his identity altered and being completely excluded left him feeling "silenced and erased" from an initiative meant to celebrate campus representation. In response to the viral backlash, Stanford officials confirmed that the alteration violated the university's strict artificial intelligence guidelines, which explicitly prohibit the creation of synthetic media depicting real individuals without consent. The university’s Residential and Dining Enterprises department has since taken down the banners and committed to implementing additional staff training. Beyond the ethical concerns of digital manipulation, the incident has reignited a heated national debate about the lengths to which academic institutions will go—including fabricating digital minorities—to project a curated image of campus diversity. source: Mallon, E. (2026). Stanford admits breaking own policy after race-, gender-swapped student photos emerge. Fox News Digital.
-
Peter Thum liked thisPeter Thum liked thisThis month, a CEO proudly told me every decision above 5,000 euros crossed his desk. He thought it proved discipline. It actually proved his culture had no autonomy left in it. By the time a customer complaint reached someone with the authority to fix it, the customer had already churned. Speed was not a technology problem in his company. It was a permission problem. Gallup's data shows highly engaged teams achieve 21% greater profitability and 17% higher productivity, and engagement rarely survives in a culture where nobody below the CEO can make a real decision. Harvard Business Review found companies with strong, healthy cultures grow revenue 4x faster than those with weak ones. Growth like that is not built in the CEO's inbox. It is built in the hundreds of small decisions employees make every day, without waiting for permission. So we changed one thing first, before touching strategy or systems. We pushed decision rights down to the person closest to the customer. A support lead could now approve a refund without 3 layers of sign-off. A product manager could now greenlight a small fix without a steering committee. Nothing about the org chart changed. What changed was who was trusted to act. Within 2 quarters, response times dropped, customer satisfaction climbed, and the CEO stopped being the bottleneck for decisions that were never his to make in the first place. Customer-centricity is not a value on a poster. It is measured in how fast the person nearest the customer is allowed to act. The only way to increase customer centricity and speed in your company culture, to 2x faster revenue growth versus your peers, is through increased autonomy of decision-making. Subscribe via the following link: https://lnkd.in/eGzVvN28 I'm Marianne van Groeningen, founder of Executive Career Coaching Europe, and post daily on leadership and human capability. We support companies in growing revenue and retaining talent by fixing root causes, not symptoms. Follow and repost along if you want strategy, people, and systems to align for 4x strategic impact.
-
Peter Thum liked thisPeter Thum liked thisMichael Burry just said OpenAI & Anthropic are using AI safety fears to strengthen their position ahead of future IPOs 😳 The investor from The Big Short isn’t convinced by calls to slow frontier AI, and his argument is all about incentives. Burry’s case has four parts: → He does not consider today’s LLMs to be AGI, so he rejects the idea that there is already an uncontrollable intelligence that needs to be slowed. → Competition is catching up quickly. Any industry-wide pause would benefit the largest incumbents by giving them more time to protect their lead. → Saying your technology may be dangerously powerful can also strengthen the perception that your models are more advanced than everyone else’s. → And if growth is slowing, safety concerns could provide a more favorable explanation for delayed launches, lower spending, or IPOs being pushed back. Of course, that doesn’t mean the risks are fake. OpenAI and Anthropic may genuinely believe frontier models are becoming harder to control. But Burry’s central point is that their commercial incentives matter too. The same companies warning that AI may be becoming dangerously powerful also benefit financially when investors believe their models are exceptionally powerful. So both things can be true: ↳ AI safety can be a legitimate concern. ↳ And safety arguments can still benefit the companies making them. If Burry is right, the number that matters next isn’t an AGI benchmark. It’s revenue growth. P.S. also check out The Agentic Transition Playbook (How founders and builders win when software starts building itself)🧠: https://lnkd.in/d-EgizPx
Experience & Education
Volunteer Experience
-
Founder and President
Foundry 47 Foundation / Liberty United Foundation
- Present 16 years 10 months
Economic Empowerment
A not for profit 501c3 organization with the mission to reduce gun violence and to support development of young people growing up at-risk of being impacted by gun violence.
-
Founder and President
Giving Water
- Present 18 years 2 months
Health
A not for profit 501c3 organization with the mission to provide safe water, sanitation and hygiene education to people in need.
To date water and sanitation have been provided to more than 6,000 schoolchildren in Kenya. -
Advisory Board Member
FEED Projects
- 5 years
Health
FEED creates good products that feed the world.
-
Board Member
Fund for Global Human Rights
- 4 years 10 months
Human Rights
The fund provides grants to front-line human rights defenders working around the world.
-
Advisory Board Member
Mgrublian Center for Human Rights at Claremont McKenna College
- 3 years 5 months
Human Rights
Instill in students an understanding of human rights as central to moral conduct and ethical decisions in their personal lives, in their careers and in the public arena. In fulfillment of its mission, the Center:
- will promote human rights globally and locally through public advocacy, service and education;
- will provide practical experience for students through internships, task forces and research; and
- will advance scholarship in the study of human rights, Holocaust history and…Instill in students an understanding of human rights as central to moral conduct and ethical decisions in their personal lives, in their careers and in the public arena. In fulfillment of its mission, the Center:
- will promote human rights globally and locally through public advocacy, service and education;
- will provide practical experience for students through internships, task forces and research; and
- will advance scholarship in the study of human rights, Holocaust history and genocide studies, including the Armenian Genocide.
-
Board Director
USA for UNHCR
- 1 year 8 months
Human Rights
USA for UNHCR is the US fundraising arm of the Office of the United Nations High Commissioner for Refugees (UNHCR), a United Nations programme mandated to protect and support refugees worldwide.
-
Advisory Board Member
NYU Robert F. Wagner Graduate School of Public Service
- 3 years 2 months
Education
NYU's Robert F. Wagner Graduate School of Public Service prepares public service leaders to translate ideas into actions that have an effective and lasting impact on the public good. Our faculty’s research changes the way people frame, understand, and act on important public issues. We provide our students with critical skills, access to all that New York City has to offer, and a deep understanding of context surrounding public service challenges—which they use to improve cities and communities…
NYU's Robert F. Wagner Graduate School of Public Service prepares public service leaders to translate ideas into actions that have an effective and lasting impact on the public good. Our faculty’s research changes the way people frame, understand, and act on important public issues. We provide our students with critical skills, access to all that New York City has to offer, and a deep understanding of context surrounding public service challenges—which they use to improve cities and communities across the globe.
-
Board Member
STARBUCKS FOUNDATION
- 3 years 1 month
The Starbucks Foundation was created as part of the Starbucks Corporation commitment to strengthen communities. It serves and invests in various social missions.
Access to Clean Water (as of the 2016 Annual Report)
Ethos Water began as a social venture startup with the goal of providing people in developing countries with access to clean water. For every bottle of Ethos® water sold in the United States, 5 cents is directed to the Ethos© Water Fund to help finance water programs…The Starbucks Foundation was created as part of the Starbucks Corporation commitment to strengthen communities. It serves and invests in various social missions.
Access to Clean Water (as of the 2016 Annual Report)
Ethos Water began as a social venture startup with the goal of providing people in developing countries with access to clean water. For every bottle of Ethos® water sold in the United States, 5 cents is directed to the Ethos© Water Fund to help finance water programs around the world. Since 2005 $13.8 million has been granted through the Ethos Water Fund, benefitting more than 500,000 people around the world. -
Member Board of Advisors
impossible2Possible
- Present 19 years 8 months
Children
The mission of impossible2Possible(i2P) is to encourage everyone to reach beyond their perceived limits, and to use adventure based learning programs to educate, inspire and empower young people all over the world.
Projects
-
Levi Strauss & Co. "Unzipped" Feature
See project"Unzipped" is the blog of Levi Strauss & Co. Here's how they describe it:
All things Levi Strauss & Co. Here, you'll find interviews with our designers, company news and exclusive looks into everything we're excited about. -
UBS "Together We Can Find an Answer" Campaign photographed by Annie Leibovitz
-
See projectAnnie Leibovitz shot a campaign for financial services company UBS, using her distinctive portraiture style to capture entrepreneurs who each represent a different dilemma such individuals might face. These form part of a wider partnership between the brand and the photographer, which has seen Leibovitz commissioned to create a new series of portraits of “women who effect positive change.” Entitled Women, the new body of work will be shown at a traveling exhibition launching in global cities…
Annie Leibovitz shot a campaign for financial services company UBS, using her distinctive portraiture style to capture entrepreneurs who each represent a different dilemma such individuals might face. These form part of a wider partnership between the brand and the photographer, which has seen Leibovitz commissioned to create a new series of portraits of “women who effect positive change.” Entitled Women, the new body of work will be shown at a traveling exhibition launching in global cities throughout 2016.
-
AT Cross "Make Your Mark" Campaign
-
In 2015, AT Cross created an advertising campaign to help people understand and reconnect with the company and the brand. Here's what the company, founded in 1846 says about its character and the concept behind this campaign:
CROSS has always championed the pursuit of greatness. From the spark that inspires a passion. To the hard work and creativity needed to make it happen. It takes courage and a lot of heart, but those who take this path are the ones who make their mark on this world.Other creatorsSee project
Honors & Awards
-
For the Love of Children Award
Children’s Home + Aid of Illinois.
Children’s Home + Aid is a leading child and family service agency in Illinois that helps children recover their health, their hope, and their faith in the people around them.
-
America's Top 50 Philanthropists
Town & Country Magazine
Recognizing men and women whose serious dollars, bold ideas, and old-fashioned hard work have made them the year's most noteworthy and inspiring givers.
-
Disruptive Innovation Award
Disruptor Foundation / Tribeca Film Festival
The Tribeca Disruptive Innovation Awards, helmed by co-founder Craig Hatkoff, in collaboration with Professor Christensen, celebrates disruptive innovation across the full spectrum of traditional and non-traditional domains ranging from technology, biomed, politics, education, healthcare, spirituality, religion, economics, sports, fashion and philanthropy.
The awards focus on breakthroughs occurring at the intersection of technology and culture where frequent clashes and resistance to…The Tribeca Disruptive Innovation Awards, helmed by co-founder Craig Hatkoff, in collaboration with Professor Christensen, celebrates disruptive innovation across the full spectrum of traditional and non-traditional domains ranging from technology, biomed, politics, education, healthcare, spirituality, religion, economics, sports, fashion and philanthropy.
The awards focus on breakthroughs occurring at the intersection of technology and culture where frequent clashes and resistance to change impede social progress and solutions for some of the world's most vexing problems. -
Eagle Scout
Boy Scouts of America
Eagle Scout is the highest achievement or rank attainable in the Boy Scouting program of the Boy Scouts of America (BSA).
Languages
-
German
Full professional proficiency
Other similar profiles
Explore more posts
Explore top content on LinkedIn
Find curated posts and insights for relevant topics all in one place.
View top content