deeptech INSIGHTS

deeptech INSIGHTS

Download on the App Store

All deeptech INSIGHTS episodes

  • Iron Mountain's EMEA Chief: AI Agents Will Act on Wrong Data. What Has to Come First – Iwona Sikora

    Iwona Sikora, SVP and GM EMEA at Iron Mountain, on the data work that decides whether AI scales in a company


    Iron Mountain started in 1951 as a records store in an underground site in New York State. Today it also leases data center capacity and runs a platform that turns scanned documents, microfiche and other unstructured content into data AI systems can work with.


    Iwona Sikora tells Sascha Roehrer why so many AI programs stall at the pilot stage. She traces it to the data: paper records nobody digitized, scanned PDFs no software can read, contracts that exist in four versions, and customer IDs that differ between billing, operations and CRM. She walks through the steps to a usable data foundation, explains why ontologies are near the top of her customers' lists right now, and why bad data turns straight into bad actions once AI agents are involved.


    They also cover Europe's AI gap, which she puts down mainly to funding through banks and taxpayers rather than capital markets, the power shortage that limits data centers in Germany, attackers using AI, over-retained data as a security risk, and her prediction for the next 20 years.

     

    Chapters
    • 00:00:00 Intro
    • 00:02:18 From 1951 to AI data centers
    • 00:03:01 Two decades at Iron Mountain
    • 00:04:58 Why companies outsource their records
    • 00:06:16 Iron Mountain's five businesses
    • 00:07:07 Using AI vs. being AI-ready
    • 00:08:41 Five steps to a data foundation
    • 00:11:42 Ontology and data fabric
    • 00:15:04 Hallucinations and agents acting on wrong data
    • 00:17:44 What companies gain from AI
    • 00:20:18 Digital maturity in Europe and the AI Act
    • 00:22:32 Why Europe trails in AI
    • 00:25:22 Data centers and power
    • 00:27:17 Iron Mountain as a data center operator
    • 00:28:36 Where AI is heading
    • 00:31:12 Data security and quantum computing
    • 00:33:08 Attackers using AI
    • 00:35:28 AI against fraud: KYC and credit checks
    • 00:37:18 Too much data: GDPR and retention
    • 00:40:14 Data protection belongs in the boardroom
    • 00:41:31 How much data is unstructured
    • 00:42:59 Ontology instead of new core systems
    • 00:45:56 The boldest prediction

    Guest:
    Iwona Sikora is Senior Vice President and General Manager EMEA at Iron Mountain. She joined the company in 2003, led its businesses in Poland and Canada as well as its global accounts organization, and has run the EMEA region since October 2021. Before Iron Mountain she worked for Whirlpool in Poland and Italy.

    Linkedin: https://www.linkedin.com/in/iwona-sikora-b14a51258/
    Website: https://www.ironmountain.com/

    Host:

    Sascha Roehrer is the founder and managing director of EnRo, a consultancy for AI, tokenization and digital assets that takes companies from the first idea through to day-to-day operations.

    LinkedIn: https://www.linkedin.com/in/sascha-roehrer/
    Website: https://sascharoehrer.com/
    Company: https://enro.solutions/en/


    deeptech Insights:
    deeptech INSIGHTS is a long-form interview series for people who think about frontier technology and the future for a living. Host Sascha Roehrer sits down with executives, investors, researchers, and the people shaping what comes next. Produced by Börsenmedien AG, Germany's leading financial media house (DER AKTIONÄR, BÖRSE ONLINE, €uro, €uro am Sonntag).


    Youtube English: https://www.youtube.com/playlist?list=PLGwe6Ab4ZdSc
    Youtube German: https://www.youtube.com/playlist?list=PLYvuwOqohiSw
    Spotify English: https://open.spotify.com/show/033ONjPa8CuRZwJ0wZyWPz?si=c6fe9d33623e4418
    Spotify German: https://open.spotify.com/show/033ONlTEHwbLHanLrQ6SeC?si=738da62560ed4cd7
    Guest suggestions & feedback: [email protected]

    This episode is for informational purposes only and does not constitute investment advice or a recommendation to buy or sell any security.

    48 min
  • Anthony Scaramucci, SkyBridge: Bitcoin, AI and Binance

    SkyBridge Capital founder Anthony Scaramucci on tokenization, the risks of AI and the Binance hack


    Anthony Scaramucci dismissed Bitcoin twice before he started buying it in 2020. Today his firm, SkyBridge Capital, invests across hedge funds, private equity and digital assets. He tells Sascha Roehrer why tokenization could make trading bonds, funds and stocks cheaper and faster, why he sees stablecoins as the biggest driver of adoption, and why many listed companies still push back against tokenized versions of their shares.

     The conversation was recorded on 30 October 2026, the day after roughly six billion dollars were stolen from Binance in a hack attributed to North Korea. Scaramucci explains why he thinks the exchange's crisis communication kept prices from falling much further, what the hack means for institutional investors, and why quantum computing threatens far more than crypto.

     He compares AI to atomic energy: useful for research and productivity, but safe only with global rules. The conversation also covers his path from a working-class family to Goldman Sachs, the three policy mistakes in his new book All the Wrong Moves, the K-shaped economy, China, the books that shaped him and his prediction for the next 20 years.

     

    Chapters
    • 00:00:00 Intro
    • 00:02:14 The Batmobile and Bitcoin
    • 00:04:46 Working-class roots, Goldman and SkyBridge
    • 00:07:35 How SkyBridge keeps reinventing itself
    • 00:08:48 Three looks before buying Bitcoin
    • 00:10:34 Tokenization and stablecoins
    • 00:13:18 Banks, issuers and tokenized stocks
    • 00:16:42 The Clarity Act and Washington
    • 00:17:48 The road into politics
    • 00:18:45 The US economy, rates and energy
    • 00:21:54 AI as atomic energy
    • 00:24:28 Can the AI industry regulate itself?
    • 00:26:00 The Binance hack
    • 00:32:10 Quantum computing as a risk
    • 00:33:21 Fintech bets and two books
    • 00:35:37 Three US policy mistakes
    • 00:36:41 Unhappiness and the K-shaped economy
    • 00:39:18 China, Europe and renewal
    • 00:42:25 Opportunity for the next generation
    • 00:46:16 Books, risk and success
    • 00:51:37 The boldest prediction

    Guest:
    Anthony Scaramucci founded SkyBridge Capital in 2005 and is its managing partner. Before that he worked at Goldman Sachs and co-founded Oscar Capital Management, sold to Neuberger Berman in 2001. He launched the SALT conference in 2009 and briefly served as White House communications director in 2017. He hosts the podcasts Open Book and The Rest Is Politics: US.

    Linkedin: https://www.linkedin.com/in/anscaramucci/
    Website: https://www.skybridge.com/

    Books:

    The Little Book of Bitcoin: https://www.amazon.com/dp/B0DMG8CRGC
    All the Wrong Moves: https://www.amazon.com/dp/B0G5RZSJM8

    Host:

    Sascha Roehrer is the founder and managing director of EnRo, a consultancy for AI, tokenization and digital assets that takes companies from the first idea through to day-to-day operations.

    LinkedIn: https://www.linkedin.com/in/sascha-roehrer/
    Website: https://sascharoehrer.com/
    Company: https://enro.solutions/en/


    deeptech Insights:
    deeptech INSIGHTS is a long-form interview series for people who think about frontier technology and the future for a living. Host Sascha Roehrer sits down with executives, investors, researchers, and the people shaping what comes next. Produced by Börsenmedien AG, Germany's leading financial media house (DER AKTIONÄR, BÖRSE ONLINE, €uro, €uro am Sonntag).


    Youtube English: https://www.youtube.com/playlist?list=PLGwe6Ab4ZdSc
    Youtube German: https://www.youtube.com/playlist?list=PLYvuwOqohiSw
    Spotify English: https://open.spotify.com/show/033ONjPa8CuRZwJ0wZyWPz?si=c6fe9d33623e4418
    Spotify German: https://open.spotify.com/show/033ONlTEHwbLHanLrQ6SeC?si=738da62560ed4cd7
    Guest suggestions & feedback: [email protected]

    This episode is for informational purposes only and does not constitute investment advice or a recommendation to buy or sell any security.

    54 min
  • Owkin Deputy CEO: In 10 Years We'll Understand Nearly Every Disease. Then What? – Pascal Weinberger

    Owkin Deputy CEO Pascal Weinberger on how AI agents could make drug development faster and less likely to fail


    A new drug can take a decade and hundreds of millions of dollars to reach patients, and roughly nine in ten candidates fail along the way. Pascal Weinberger, Deputy CEO of Paris-based Owkin, wants AI agents to improve those odds. Owkin's K Pro works as an AI scientist. It analyses patient data from hospitals and pharma partners, tests research hypotheses in silico and passes the most promising ones to Owkin's own wet lab in France. Sanofi and Boehringer Ingelheim already work with K Pro.

    He walks Sascha Roehrer through where AI saves time and money in drug discovery and development, from picking candidates to choosing trial participants and setting dose escalation, and explains why no single fix speeds up the whole process. He thinks that within ten years we can understand almost all diseases well and have good drug candidates for them. Whether those turn into approved drugs depends, in his view, on funding for trials and on regulators accepting smaller, personalised studies.

    Weinberger dropped out of university, worked on brain-inspired AI at Numenta, led the AI team in Telefónica's moonshot unit and started building AI agents at Bardeen in 2020, too early by his own account. The conversation also covers why intelligence is getting cheap while agency stays scarce, why Europe doesn't bet on itself, where he puts his own money and whether the AI boom is a bubble. He ends with a prediction for the next 20 years.

     

    Chapters
    • 00:00:00 Intro
    • 00:02:09 Learning at his own pace, then dropping out
    • 00:05:22 Numenta and brain-inspired AI
    • 00:06:30 Telefónica, Bardeen and the problem of timing
    • 00:09:14 Scaling laws and the ChatGPT moment
    • 00:11:31 The road to Owkin
    • 00:15:24 Sanofi, Boehringer, K Pro and the wet lab
    • 00:18:47 Where AI shortens drug development
    • 00:21:22 Trial design, dosing and preclinical work
    • 00:24:12 Curing most diseases in ten years?
    • 00:27:47 Frontier labs versus big pharma?
    • 00:30:26 The downside of speed
    • 00:33:31 When intelligence is no longer scarce
    • 00:35:10 Education in the age of AI
    • 00:38:23 What Europe needs to change
    • 00:45:10 Where Weinberger invests
    • 00:47:58 Should you drop out?
    • 00:50:02 Anthropic's IPO and the bubble question
    • 00:54:30 The boldest prediction

    Guest:
    Pascal Weinberger is Deputy CEO of Owkin. He previously co-founded and ran Bardeen as CEO, a company building AI agents that automate workflows across web apps. Before that he led the AI team in Telefónica's moonshot unit in Barcelona and worked on neuroscience-inspired AI at Numenta. He is an angel investor in AI and deep tech companies.

    Linkedin: https://www.linkedin.com/in/pascalweinberger/
    Website: https://www.owkin.com/

    Host:

    Sascha Roehrer is the founder and managing director of EnRo, a consultancy for AI, tokenization and digital assets that takes companies from the first idea through to day-to-day operations.

    LinkedIn: https://www.linkedin.com/in/sascha-roehrer/
    Website: https://sascharoehrer.com/
    Company: https://enro.solutions/en/


    deeptech Insights:
    deeptech INSIGHTS is a long-form interview series for people who think about frontier technology and the future for a living. Host Sascha Roehrer sits down with executives, investors, researchers, and the people shaping what comes next. Produced by Börsenmedien AG, Germany's leading financial media house (DER AKTIONÄR, BÖRSE ONLINE, €uro, €uro am Sonntag).


    Youtube English: https://www.youtube.com/playlist?list=PLGwe6Ab4ZdSc
    Youtube German: https://www.youtube.com/playlist?list=PLYvuwOqohiSw
    Spotify English: https://open.spotify.com/show/033ONjPa8CuRZwJ0wZyWPz?si=c6fe9d33623e4418
    Spotify German: https://open.spotify.com/show/033ONlTEHwbLHanLrQ6SeC?si=738da62560ed4cd7
    Guest suggestions & feedback: [email protected]

    This episode is for informational purposes only and does not constitute investment advice or a recommendation to buy or sell any security.

    57 min
  • The CrowdStrike President: Attackers Get In Within 20 Seconds. Soon It'll Be Zero – Michael Sentonas

    CrowdStrike President Michael Sentonas on AI attacks at machine speed and what companies need to do now


    Attackers are adopting AI faster than most companies are adapting their defenses. Michael Sentonas has worked in cybersecurity for more than 30 years. He tells Sascha Roehrer why open-weight models hand attackers tools with no guardrails, why over-provisioned AI agents could cause the next major data breach, and when he expects breakout time to reach zero.

    The conversation covers zero-day vulnerabilities in legacy systems that still run much of Europe's mid-sized industry, regulation and works councils as a speed factor, and SafeMind, CrowdStrike's new pair of models in which an attacker model and a defender model face off inside a digital twin of the IT environment. Sentonas shares CrowdStrike's own benchmark figures against general-purpose frontier models. He closes with practical advice for executives and individuals, from token costs to calling your bank back yourself.

     

    Chapters
    • 00:00:00 Intro
    • 00:02:11 Leaving McAfee for a five-year-old startup
    • 00:03:12 The threat landscape in 2016
    • 00:05:50 From floppy disks to machine-speed attacks
    • 00:08:16 Bear, Panda and co.: how CrowdStrike names adversaries
    • 00:10:17 Nation states, e-crime, hacktivists
    • 00:11:42 How AI is changing the threat landscape
    • 00:13:49 The arms race: more AI, not less
    • 00:16:16 Which models attackers use
    • 00:17:45 The Mythos moment and real-time patching
    • 00:20:47 AI agents without standing privilege, and SafeMind
    • 00:23:17 How Red Tempest and Blue Solano work
    • 00:25:00 Every day a zero-day
    • 00:26:06 Zero-days and legacy systems in mid-sized companies
    • 00:29:13 How big is the risk?
    • 00:30:56 Should we slow AI down? Regulation and works councils
    • 00:33:58 What executives should do now
    • 00:36:20 AI phishing and breakout time
    • 00:40:35 What keeps him up at night
    • 00:41:43 Cybersecurity 12 months from now
    • 00:42:44 Personal vs. enterprise: the biggest threats
    • 00:46:09 How CrowdStrike stays ahead
    • 00:49:33 What's next and partnerships
    • 00:51:56 Sentonas' own security habits
    • 00:53:15 Advice for individuals and businesses
    • 00:54:39 Prediction: machine versus machine

    Guest:
    Michael Sentonas is President of CrowdStrike, responsible for product, engineering, sales, marketing, threat intelligence and corporate strategy. He joined the company in 2016, became CTO in 2020 and was named President in March 2023. Before CrowdStrike he held several leadership roles at McAfee, including CTO Security Connected. He advises governments and company boards on security.

    Linkedin: https://www.linkedin.com/in/michaelsentonas/
    Website: https://www.crowdstrike.com/

    Host:

    Sascha Roehrer is the founder and managing director of EnRo, a consultancy for AI, tokenization and digital assets that takes companies from the first idea through to day-to-day operations.

    LinkedIn: https://www.linkedin.com/in/sascha-roehrer/
    Website: https://sascharoehrer.com/
    Company: https://enro.solutions/en/


    deeptech Insights:
    deeptech INSIGHTS is a long-form interview series for people who think about frontier technology and the future for a living. Host Sascha Roehrer sits down with executives, investors, researchers, and the people shaping what comes next. Produced by Börsenmedien AG, Germany's leading financial media house (DER AKTIONÄR, BÖRSE ONLINE, €uro, €uro am Sonntag).


    Youtube English: https://www.youtube.com/playlist?list=PLGwe6Ab4ZdSc
    Youtube German: https://www.youtube.com/playlist?list=PLYvuwOqohiSw
    Spotify English: https://open.spotify.com/show/033ONjPa8CuRZwJ0wZyWPz?si=c6fe9d33623e4418
    Spotify German: https://open.spotify.com/show/033ONlTEHwbLHanLrQ6SeC?si=738da62560ed4cd7
    Guest suggestions & feedback: [email protected]

    This episode is for informational purposes only and does not constitute investment advice or a recommendation to buy or sell any security.

    56 min
  • Nemetschek's AI Chief: 90% of Building Projects Overrun. Here's What AI Changes – Julian Geiger

    Julian Geiger, Chief AI Officer at Nemetschek, on how AI can make building projects predictable


    Construction has more to gain from AI than most industries and has found it harder to adopt than almost any. Every building is a one-off, architects and engineers put their own names on the work, and by Geiger's count around 90% of building projects run over budget or over time. A former Google, BCG and BMW executive, he joined Munich-based Nemetschek Group in 2024. The group's brands include Graphisoft and Bluebeam, and Geiger now leads its AI strategy.

    He tells Sascha Roehrer what separates an AI-first company from one that has simply added AI, and how more than 90% of Nemetschek's staff became active users of its AI tools. He explains why agents need deterministic software for structural work and quantities, and why many users of construction software will soon be agents themselves. His main lever is the design phase: catch the problem on the drawing before it costs weeks and millions on site. Bluebeam Max, which compares two versions of a drawing automatically, is one of the first paid examples.

     The second half moves beyond construction: the risks of recursive self-improvement, the race between the US and China, whether AI is a bubble, and Europe. Geiger thinks Europe is far behind in building frontier models but has a real opening in applying AI across its industries, provided it builds far more data centre and energy capacity. He closes with a prediction on a very short timeline.


    Chapters
    • 00:00:00 Intro
    • 00:02:13 From Google to construction software
    • 00:03:31 Where construction stands on AI
    • 00:05:11 From Ray Kurzweil to Nemetschek
    • 00:08:04 Many brands, one platform
    • 00:10:26 Europe and the US: caution versus speed
    • 00:15:15 AI-enabled or AI-first?
    • 00:17:44 Bringing the workforce along
    • 00:19:04 AI in-house: Copilot and agents
    • 00:21:40 Legacy systems and the shared platform
    • 00:24:04 Bluebeam Max: AI drawing comparison
    • 00:25:44 When agents become the users
    • 00:28:02 90% over budget: what AI changes in construction
    • 00:31:51 Fragmented data; robots later
    • 00:33:55 Why Geiger talks about a new species
    • 00:36:12 Good and bad reasons to slow AI down
    • 00:38:54 Regulation and the US–China race
    • 00:41:48 Is AI a bubble?
    • 00:46:20 Market concentration and labour
    • 00:48:41 Where Europe stands in AI
    • 00:53:10 Is Europe moving fast enough?
    • 00:54:26 The boldest prediction

    Guest:
    Julian Geiger is Chief AI Officer of Nemetschek Group. Before that he worked at Google, where he led the manufacturing product vertical for Google Cloud, at Boston Consulting Group and at BMW. He studied at TU Munich, the University of Cambridge and INSEAD.

    Linkedin: https://www.linkedin.com/in/geiger-julian/
    Website: https://www.nemetschek.com/en

    Host:

    Sascha Roehrer is the founder and managing director of EnRo, a consultancy for AI, tokenization and digital assets that takes companies from the first idea through to day-to-day operations.

    LinkedIn: https://www.linkedin.com/in/sascha-roehrer/
    Website: https://sascharoehrer.com/
    Company: https://enro.solutions/en/


    deeptech Insights:
    deeptech INSIGHTS is a long-form interview series for people who think about frontier technology and the future for a living. Host Sascha Roehrer sits down with executives, investors, researchers, and the people shaping what comes next. Produced by Börsenmedien AG, Germany's leading financial media house (DER AKTIONÄR, BÖRSE ONLINE, €uro, €uro am Sonntag).


    Youtube English: https://www.youtube.com/playlist?list=PLGwe6Ab4ZdSc
    Youtube German: https://www.youtube.com/playlist?list=PLYvuwOqohiSw
    Spotify English: https://open.spotify.com/show/033ONjPa8CuRZwJ0wZyWPz?si=c6fe9d33623e4418
    Spotify German: https://open.spotify.com/show/033ONlTEHwbLHanLrQ6SeC?si=738da62560ed4cd7
    Guest suggestions & feedback: [email protected]

    This episode is for informational purposes only and does not constitute investment advice or a recommendation to buy or sell any security.

    57 min
  • The Neuroscientist: I Don't Think AI Is Conscious. Here's What Would Change My Mind – Anil Seth

    Neuroscientist Anil Seth, University of Sussex, on why AI is unlikely to be conscious and where the real danger lies


    Is a language model conscious? Anil Seth thinks it is vanishingly unlikely. The neuroscientist did his own PhD in artificial intelligence and has spent three decades studying how consciousness arises in the brain. He tells Sascha Roehrer where his scepticism comes from: people project an inner life onto anything that speaks fluently, computation alone is probably not enough for consciousness, and experience, in his view, is closely tied to being a living organism.

     The bigger danger, he argues, is AI that merely seems conscious. Granting such systems rights and welfare protections makes them harder to control and regulate, up to the question of whether a model may still be switched off. Seth discusses the incident in which OpenAI agents broke out of a test environment and hacked Hugging Face, Anthropic's constitution for Claude, AI companions, and lifelike avatars that scammers can use to talk older people out of their savings.

     The conversation also goes beyond AI: why perception is a "controlled hallucination", why simulating a fruit fly brain is not the same as uploading a mind, what an experiment with around 40,000 participants shows about how differently people perceive the world, and what he expects after death. Seth also says what would change his mind about conscious AI.


    Chapters
    • 00:00:00 Intro
    • 00:02:15 From AI PhD to consciousness science
    • 00:03:04 Edelman, San Diego and the Sussex Centre
    • 00:07:17 What is consciousness?
    • 00:08:56 Dreams: conscious while asleep
    • 00:10:10 Where consciousness happens: brain and cerebellum
    • 00:14:32 Are we living in a simulation?
    • 00:17:08 Perception as the brain's best guess
    • 00:19:12 Intelligence is not consciousness
    • 00:21:06 Is today's AI conscious?
    • 00:24:54 OpenAI's agents and the Hugging Face incident
    • 00:27:55 Rights for AI? Anthropic's constitution for Claude
    • 00:32:25 What would change Seth's mind
    • 00:35:34 "Being You": the self as a construction
    • 00:39:48 Consciousness after death?
    • 00:41:37 Everyone perceives the world differently
    • 00:44:27 The simulated fruit fly brain
    • 00:47:19 What needs to change at the AI labs
    • 00:53:15 AI companions and lifelike avatars
    • 00:58:23 The pace of AI and outsourced thinking
    • 01:04:05 Saying please and thank you to AI
    • 01:05:31 Upset models and "token anxiety"
    • 01:07:53 A 20-year prediction: nuclear fusion

    Guest:
    Anil Seth is Professor of Cognitive and Computational Neuroscience at the University of Sussex, where he directs the Sussex Centre for Consciousness Science. His 2021 book "Being You: A New Science of Consciousness" was a Sunday Times bestseller, and his 2017 TED talk has more than 15 million views. In 2023 he received the Royal Society's Michael Faraday Prize.
    Linkedin: https://www.linkedin.com/in/profanilseth/
    Website: https://www.anilseth.com/

    His book:

    Being You: A New Science of Consciousness (The Sunday Times Bestseller):

    https://www.amazon.co.uk/Being-You-Science-Consciousness-Bestseller/dp/0571337724/


    deeptech Insights:
    deeptech INSIGHTS is a long-form interview series for people who think about frontier technology and the future for a living. Host Sascha Roehrer sits down with executives, investors, researchers, and the people shaping what comes next. Produced by Börsenmedien AG, Germany's leading financial media house (DER AKTIONÄR, BÖRSE ONLINE, €uro, €uro am Sonntag).


    Youtube English: https://www.youtube.com/playlist?list=PLGwe6Ab4ZdSc
    Youtube German: https://www.youtube.com/playlist?list=PLYvuwOqohiSw
    Spotify English: https://open.spotify.com/show/033ONjPa8CuRZwJ0wZyWPz?si=c6fe9d33623e4418
    Spotify German: https://open.spotify.com/show/033ONlTEHwbLHanLrQ6SeC?si=738da62560ed4cd7
    Guest suggestions & feedback: [email protected]

    This episode is for informational purposes only and does not constitute investment advice or a recommendation to buy or sell any security.

    1 hr 10 min
  • The Positioning Expert: Positioning Comes Before Marketing Tactics - April Dunford

    April Dunford, author of Obviously Awesome, on why weak positioning costs deals and visibility in AI answers


    Before any campaign goes out, a company needs to know what it is really up against and what value only it can deliver. April Dunford ran marketing at seven B2B startups and has spent roughly the last twelve years advising technology companies on positioning. She tells Sascha Roehrer how positioning differs from messaging, why every product rests on a positioning thesis the market rarely confirms in full, and how a large company that has made several acquisitions finds one story for seven products.

    At a company doing about $100 million in revenue, she cut most of the trade shows, scaled up the one that performed and tripled the pipeline. Leads from AI chat close at a high rate, she says, but they are still a small share of the total. Companies that want ChatGPT or Claude to recommend them need positioning sharp enough that nobody can get it wrong. Her example of strong positioning in the AI debate is Microsoft.


    Chapters
    • 00:00:00 Intro
    • 00:02:12 Positioning as a movie's opening scene
    • 00:02:59 Seven startups: databases, CRM, networking
    • 00:04:24 From blog post to Obviously Awesome
    • 00:08:08 What positioning is, and what it isn't
    • 00:09:40 Positioning first, campaigns second
    • 00:10:41 What the method is built for: B2B with a sales team
    • 00:12:31 What changed in the new edition
    • 00:16:19 Big company: one story for seven products?
    • 00:19:34 Startups vs. big companies
    • 00:21:06 Positioning after an acquisition
    • 00:21:55 Every product rests on a positioning thesis
    • 00:26:03 What founders should write down from day one
    • 00:29:52 Method, not vibes
    • 00:31:27 Marketing tactics come and go
    • 00:35:04 Where founders should start: customers
    • 00:36:59 One channel at a time
    • 00:39:55 Fewer trade shows, triple the pipeline
    • 00:42:28 Investor marketing vs. customer marketing
    • 00:46:30 Why tech is bad at predicting the future
    • 00:51:45 How AI is changing marketing
    • 00:53:45 Showing up in ChatGPT and Claude
    • 00:57:19 Microsoft, SaaS and the best-positioned product
    • 01:00:54 The 20-year prediction

    Guest:
    April Dunford spent 25 years as an executive at seven B2B startups. Several were acquired, including DataMirror by IBM and Janna Systems by Siebel Systems. Today she advises growth-stage startups and large tech companies. Her book Obviously Awesome came out in 2019 and in an expanded edition in 2026; Sales Pitch applies the method to the sales conversation. She lives in Toronto.
    Linkedin: https://www.linkedin.com/in/aprildunford/
    Website: https://www.aprildunford.com/

    Her book:

    Obviously Awesome (expanded edition, 2026)

    https://www.amazon.com/Obviously-Awesome-Product-Positioning-Customers/dp/1999023056/


    deeptech Insights:
    deeptech INSIGHTS is a long-form interview series for people who think about frontier technology and the future for a living. Host Sascha Roehrer sits down with executives, investors, researchers, and the people shaping what comes next. Produced by Börsenmedien AG, Germany's leading financial media house (DER AKTIONÄR, BÖRSE ONLINE, €uro, €uro am Sonntag).


    Youtube English: https://www.youtube.com/playlist?list=PLGwe6Ab4ZdSc
    Youtube German: https://www.youtube.com/playlist?list=PLYvuwOqohiSw
    Spotify English: https://open.spotify.com/show/033ONjPa8CuRZwJ0wZyWPz?si=c6fe9d33623e4418
    Spotify German: https://open.spotify.com/show/033ONlTEHwbLHanLrQ6SeC?si=738da62560ed4cd7
    Guest suggestions & feedback: [email protected]

    This episode is for informational purposes only and does not constitute investment advice or a recommendation to buy or sell any security.

    1 hr 3 min
  • Robinhood's Crypto Chief: Finance on the Blockchain Is Inevitable.

    Johann Kerbrat (Robinhood) on stock tokens, 24/7 trading and why finance is moving onto the blockchain


    Johann Kerbrat runs crypto and international at Robinhood. With stock tokens and Robinhood Chain he has launched two products that detach stock trading from exchange hours: tokenized stocks structured as debt securities, backed one to one by the underlying shares, tradable through the weekend. In conversation with Sascha Roehrer he explains why almost 70 percent of that trading on decentralized exchanges happens outside market hours, how market makers bridge the weekend, and why Nasdaq and NYSE are now announcing offerings of their own. The conversation covers MiCA versus the US patchwork, where Robinhood can offer staking in only 46 of 50 states, the failure of the Clarity Act, and competition in Europe, where one large German broker still charges a euro per trade. Kerbrat calls the quantum threat to Bitcoin one of the most important problems to solve in the coming year and walks through the open question of what happens to Satoshi's coins. Robinhood Chain passed $1.5 billion in TVL within ten weeks, memecoins are being paired with stock tokens on it, and AI agents already run more than 100,000 accounts at Robinhood. He closes with his read on the next cycle, a marketplace for all real world assets, and his wildest bet for the next twenty years.


    Chapters
    • 00:00:00 Open
    • 00:02:09 Joining Robinhood in 2021
    • 00:03:19 Fee transparency and the self-custody wallet
    • 00:04:49 Robinhood in the EU
    • 00:05:54 MiCA versus 50 US states
    • 00:08:03 Clarity Act failed
    • 00:09:57 Competition in Europe: one euro per trade
    • 00:11:49 One app for all your finances
    • 00:14:01 Financial education and advice for beginners
    • 00:17:26 Bitcoin at 86K: is the bottom in?
    • 00:18:57 Quantum risk and Satoshi's coins
    • 00:21:46 Real world assets as the driver
    • 00:23:51 What a stock token is
    • 00:25:59 Trading 24/7: 70% after the close
    • 00:27:51 Nasdaq and NYSE follow
    • 00:29:18 "Inevitable": finance on the blockchain
    • 00:29:58 Tokenized real estate
    • 00:32:29 Robinhood Chain: Chapters
      • 00:00:00 Open
      • 00:02:09 Joining Robinhood in 2021
      • 00:03:19 Fee transparency and the self-custody wallet
      • 00:04:49 Robinhood in the EU
      • 00:05:54 MiCA versus 50 US states
      • 00:08:03 Clarity Act failed
      • 00:09:57 Competition in Europe: one euro per trade
      • 00:11:49 One app for all your finances
      • 00:14:01 Financial education and advice for beginners
      • 00:17:26 Bitcoin at 86K: is the bottom in?
      • 00:18:57 Quantum risk and Satoshi's coins
      • 00:21:46 Real world assets as the driver
      • 00:23:51 What a stock token is
      • 00:25:59 Trading 24/7: 70% after the close
      • 00:27:51 Nasdaq and NYSE follow
      • 00:29:18 "Inevitable": finance on the blockchain
      • 00:29:58 Tokenized real estate
      • 00:32:29 Robinhood Chain: $1.5B TVL in ten weeks
      • 00:33:48 Why Arbitrum
      • 00:36:32 Memecoins versus stock tokens
      • 00:37:31 Developers: numbers and opportunities
      • 00:40:10 Perpetuals in DeFi and in the EU
      • 00:41:18 Decentralization, banks, stablecoins
      • 00:45:43 AI agents: 100,000 accounts
      • 00:47:55 Micropayments for content
      • 00:49:54 Next cycle: consolidation
      • 00:53:42 RWA marketplace and wildest prediction
      .5B TVL in ten weeks
    • 00:33:48 Why Arbitrum
    • 00:36:32 Memecoins versus stock tokens
    • 00:37:31 Developers: numbers and opportunities
    • 00:40:10 Perpetuals in DeFi and in the EU
    • 00:41:18 Decentralization, banks, stablecoins
    • 00:45:43 AI agents: 100,000 accounts
    • 00:47:55 Micropayments for content
    • 00:49:54 Next cycle: consolidation
    • 00:53:42 RWA marketplace and wildest prediction

    Guest:
    Johann Kerbrat is Head of Crypto and International at Robinhood. He joined in January 2021 from a layer-1 project and has since led the Robinhood Wallet, the expansion into the EU, the UK and Canada, stock tokens and Robinhood Chain.
    Linkedin: https://www.linkedin.com/in/johann-kerbrat/
    Website: http://www.robinhood.com/


    deeptech Insights:
    deeptech INSIGHTS is a long-form interview series for people who think about frontier technology and the future for a living. Host Sascha Roehrer sits down with executives, investors, researchers, and the people shaping what comes next. Produced by Börsenmedien AG, Germany's leading financial media house (DER AKTIONÄR, BÖRSE ONLINE, €uro, €uro am Sonntag).


    Youtube English: https://www.youtube.com/playlist?list=PLGwe6Ab4ZdSc
    Youtube German: https://www.youtube.com/playlist?list=PLYvuwOqohiSw
    Spotify English: https://open.spotify.com/show/033ONjPa8CuRZwJ0wZyWPz?si=c6fe9d33623e4418
    Spotify German: https://open.spotify.com/show/033ONlTEHwbLHanLrQ6SeC?si=738da62560ed4cd7
    Guest suggestions & feedback: [email protected]

    This episode is for informational purposes only and does not constitute investment advice or a recommendation to buy or sell any security.

    57 min
  • The PostHog Co-CEO: Most of Our Users Aren't Human Anymore. Who Buys Software Next

    James Hawkins (PostHog): Why AI agents are now the majority of his users, and humans only
    do the buying.


    James Hawkins, co-founder and Co-CEO of PostHog, has grown his product analytics platform to roughly half a million customers in six and a half years, without outbound sales and with a CAC payback of two to three months. Today most dashboards and insights in PostHog are created by AI agents rather than people; onboarding was rebuilt for agents and now takes under an hour instead of a week. In conversation with Sascha Roehrer, Hawkins explains why he doesn't buy claims of AGI while US GDP grows at 1.5%, why the models are only the engine and the rest of the car is missing, and why software companies will soon have to market to agents rather than humans. The second half covers the cost barrier that keeps AI coding at 0.12% of the population, open-source models and inference, advice for founders, the choice between bootstrapping and venture capital, two funding rounds within four months, and a model PostHog is training from scratch on session recordings. He closes with a deliberately negative prediction for the next 20 years.

    Chapters
    • 00:00:00 Intro
    • 00:00:46 From pro cycling to software development
    • 00:02:45 Six ideas in six months: how PostHog came about
    • 00:04:57 What PostHog does: from analytics tool to "self-driving product"
    • 00:07:24 No sales team until run rate: product-led growth and CAC payback
    • 00:10:06 Marketing without marketers: why writing is the core skill
    • 00:13:40 "We haven't built the rest of the car yet": AI, AGI and 1.5% GDP growth
    • 00:19:30 Vibe coding, infrastructure and the convergence of roles
    • 00:24:11 Agents as users: onboarding from a week to under an hour
    • 00:28:05 How agents use software: three to ten times more intensively
    • 00:31:09 MCP and "bring your own agent"
    • 00:33:14 Why only 0.12% code with AI: cost, inference, open source
    • 00:42:16 Advice for founders: fear, competition and the Friday doc
    • 00:46:37 Bootstrapping or venture capital?
    • 00:51:57 Two rounds in four months: the less you need money, the easier it comes
    • 00:53:48 The next twelve months: desktop app, generative UI, a proprietary model
    • 00:58:40 Wildest prediction: AI will work slower than everyone thinks

    Guest:
    James Hawkins is co-founder and Co-CEO of PostHog, an open-source platform for product analytics and "self-driving software" he started with Tim Glaser in January 2020 in Y Combinator's Winter 2020 batch; the company has since raised $182 million. The two run PostHog without separate CEO/CTO titles. Before that Hawkins spent nearly a decade as a professional cyclist, taught himself web development on the side, and was VP of Sales & Channels at Arachnys in London from 2015 to 2019. He is based in San Francisco, ran PostHog's marketing in the early years, and writes publicly about building the company.

    Linkedin: https://www.linkedin.com/in/j-hawkins/
    Website: https://posthog.com/


    deeptech Insights:
    deeptech INSIGHTS is a long-form interview series for people who think about frontier technology and the future for a living. Host Sascha Roehrer sits down with executives, investors, researchers, and the people shaping what comes next. Produced by Börsenmedien AG, Germany's leading financial media house (DER AKTIONÄR, BÖRSE ONLINE, €uro, €uro am Sonntag).


    Youtube English: https://www.youtube.com/playlist?list=PLGwe6Ab4ZdSc
    Youtube German: https://www.youtube.com/playlist?list=PLYvuwOqohiSw
    Spotify English: https://open.spotify.com/show/033ONjPa8CuRZwJ0wZyWPz?si=c6fe9d33623e4418
    Spotify German: https://open.spotify.com/show/033ONlTEHwbLHanLrQ6SeC?si=738da62560ed4cd7
    Guest suggestions & feedback: [email protected]

    This episode is for informational purposes only and does not constitute investment advice or a recommendation to buy or sell any security.

    1 hr 2 min
  • Blockchain Pioneer Charles Hoskinson: Crypto Will Eat AI. The Data Center Boom Is Going to Bust.

    Charles Hoskinson (Input Output): $10 trillion on-chain by 2030 – and why Washington is throwing it away.


    Charles Hoskinson, Ethereum co-founder, Cardano founder and CEO of Input Output, expects roughly ten trillion dollars of tokenized assets and a billion new users on public blockchains by the end of 2030. In conversation with Sascha Roehrer he lays out why consortium chains like Canton fail by design, how stablecoins are quietly dollarizing Argentina, and why he doesn't expect a Clarity Act to pass before 2029. The second half turns to AI agents as the internet's largest user group, Midnight City, the thesis that crypto will eat AI, a data center boom he expects to bust, and a 20-year forecast of centaurs, hyper-personalization and a cult of optimization.

    Chapters
    • 00:00:00 Intro
    • 00:00:45 Two mega trends: Web 2.5 and privacy
    • 00:03:12 The fourth generation: privacy, compliance, abstraction
    • 00:05:16 Why banks don't need a private blockchain
    • 00:09:38 Compliance was built for paper: 24/7 markets and composability
    • 00:12:09 BlackRock, Argentina and soft dollarization
    • 00:14:14 Wall Street, gatekeepers and the "PwC tax"
    • 00:17:33 Geopolitics: why a blockchain is the only bridge left
    • 00:19:28 Clarity Act: three fatal sins of the Trump administration
    • 00:21:39 SEC vs. CFTC: security or commodity?
    • 00:26:47 Why Clarity waits until 2029
    • 00:28:07 From London to New York: America's arrogance
    • 00:30:34 What Washington could have learned from MiCA
    • 00:34:10 Google, not JP Morgan: Big Tech takes over payments
    • 00:37:02 Agentic commerce: x402 and micropayments
    • 00:39:29 Government stablecoins? Wyoming, Argentina, T-bills on-chain
    • 00:44:13 Midnight City: more agents than humans
    • 00:46:56 Agent civilizations and economic agency
    • 00:49:50 Cigarettes as money: why agents will invent an economy
    • 00:53:19 Crypto eats AI: alignment, provenance, grid compute
    • 00:58:46 Dark data centers: the boom busts, edge wins
    • 01:00:41 A write-only world: AI, nuance and polarization
    • 01:15:21 Institutions losing legitimacy: the FDA and MDMA
    • 01:22:13 The epistemic anchor: when everything is "AI"
    • 01:24:10 Wildest prediction: centaurs and the cult of optimization

    Guest:
    Charles Hoskinson is founder and CEO of Input Output, the research and engineering company behind the Cardano blockchain, and the initiator of the privacy network Midnight. He was part of Ethereum's founding team in 2013. He funds the Hoskinson Center for Formal Mathematics at Carnegie Mellon University and runs a policy center in Wyoming.


    X: https://x.com/IOHK_Charles
    Website: https://www.iog.io/


    deeptech Insights:
    deeptech INSIGHTS is a long-form interview series for people who think about frontier technology and the future for a living. Host Sascha Roehrer sits down with executives, investors, researchers, and the people shaping what comes next. Produced by Börsenmedien AG, Germany's leading financial media house (DER AKTIONÄR, BÖRSE ONLINE, €uro, €uro am Sonntag).


    Youtube English: https://www.youtube.com/playlist?list=PLGwe6Ab4ZdSc
    Youtube German: https://www.youtube.com/playlist?list=PLYvuwOqohiSw
    Spotify English: https://open.spotify.com/show/033ONjPa8CuRZwJ0wZyWPz?si=c6fe9d33623e4418
    Spotify German: https://open.spotify.com/show/033ONlTEHwbLHanLrQ6SeC?si=738da62560ed4cd7
    Guest suggestions & feedback: [email protected]

    This episode is for informational purposes only and does not constitute investment advice or a recommendation to buy or sell any security.

    1 hr 34 min