Builders by Proxify

In Silicon Valley, failure is often seen as valuable experience. Risk-taking is encouraged, ambitious ideas attract investment, and companies are expected to move fast. But what can Sweden and the Nordics learn from this approach, and what strengths do they already have?In this episode of Builders, our host Lee Haynes speaks with Joakim Skog, Head of Office and Strategy at Vinnova, Sweden’s innovation agency, based in Silicon Valley.They discuss the differences between Swedish and American innovation cultures, why Silicon Valley rewards failure, how governments can support ambitious technological breakthroughs, and what it takes to build a successful innovation ecosystem.They also explore strategic foresight, why predicting the future isn’t the goal, and how organizations can prepare for technological changes they can’t yet fully anticipate.In this episode:• Why Silicon Valley treats failure differently• What Sweden can learn from the US innovation ecosystem• How public funding supports high-risk innovation• Why Sweden needs to make strategic bets on emerging technologies• How strategic foresight helps organizations prepare for the future• How to distinguish meaningful signals of change from passing trendsThis is Part 1 of our conversation with Joakim Skog. In Part 2, we explore emerging technologies, robotics, neurotechnology, and Europe’s position in the global technology race.Subscribe to Builders for conversations with the people shaping the future of technology, engineering, and innovation.Chapters:00:00 Introduction: What can Sweden learn from Silicon Valley?01:08 Why Vinnova is based in Silicon Valley03:33 What makes the US innovation ecosystem different?05:30 Why Silicon Valley rewards failure06:56 Combining the best of Sweden and Silicon Valley10:22 The DARPA model and ambitious moonshots11:44 The role of public funding in innovation15:00 How Sweden chooses which technologies to invest in18:41 Strategic foresight: Can we predict the future?21:45 Which emerging trends are leaders underestimating?24:26 How to separate meaningful signals from noise

Are we building AI because we need it, or because we can?As companies invest billions in larger AI models, more computing power, and new data centers, an important question remains: what are we building all of this for?

In Part 2 of Builders, our host Lee Haynes continues his conversation with Charlotta Brask, Head of Sustainability at Telia Sweden, about the human side of sustainable AI and the responsibility companies have when developing and adopting new technologies.

With over 20 years of experience in sustainability, Charlotta explains why responsible AI requires more than technical expertise. It also means considering human rights, involving different perspectives in development decisions, and questioning whether bigger AI systems are always the right solution.

They discuss the influence of big tech on AI development, the importance of involving sustainability professionals in AI governance, and practical ways engineering teams can reduce the environmental impact of their systems.In this episode:

• The human and social impact of AI development
• Why sustainability professionals need a seat at the table
• The influence of big tech on the future of AI
• Why bigger AI models aren’t always better
• Practical approaches to building more sustainable AI systems
• What responsible AI development should look like in the coming years

Chapters:

00:00 The human side of sustainable AI

03:25 Are companies becoming more aware of AI’s impact?

05:23 Why bigger AI isn’t always better

07:41 Practical advice for building sustainable AI

10:41 What needs to change in the next three years?

12:20 Final thoughts on responsible AI

This is Part 2 of our conversation with Charlotta Brask. Watch Part 1    • The hidden environmental cost of AI, with ...   to explore the environmental footprint of AI, from energy and water consumption to hardware production and the global supply chain.

Subscribe to Builders for more conversations with the people building the future of technology.

AI promises to help build a more sustainable future. But how sustainable is AI itself?Behind every AI model is a physical infrastructure that consumes energy, water, and natural resources. And as companies race to adopt AI, its environmental footprint is becoming harder to ignore.In this episode of Builders, Lee Haynes sits down with Charlotta Brask, Head of Sustainability at Telia Sweden, to discuss the hidden environmental costs of AI and what companies can do to build more responsibly.With over 20 years of experience in sustainability, Charlotta shares her perspective on why bigger AI models aren’t always better, how companies can measure their AI footprint, and why sustainability needs to become part of everyday engineering decisions.They discuss the environmental impact of data centers, the resources required to manufacture AI hardware, and how Telia is integrating sustainability into its AI governance.In this episode:

• The difference between AI for sustainability and sustainable AI
• The hidden environmental and social costs of AI
• Why smaller AI models can be a smarter choice
• How to measure energy, water, and resource consumption
• The role of engineering teams in building sustainable AI
• How Telia approaches responsible AI development

Can AI help solve environmental challenges without creating new ones?Subscribe to Builders for more conversations with the people building the future of technology.

Are we losing control of AI? As intelligent systems become increasingly autonomous, who takes responsibility for the decisions they make?In Part 2 of Builders, our host Lee Haynes continues his conversation with Ingo Paas, former CIO of Green Cargo, now board member at Svenska kraftnät, and author of A Billion Times Smarter and Digital Composable

Enterprises.Drawing on his experience leading digital transformation and researching the implications of superintelligence, Ingo explores what happens when decision-making authority shifts from humans to intelligent systems.They discuss what it means to preserve human agency in an AI-driven world, whether we can maintain control over increasingly autonomous systems, and why traditional approaches to AI governance may no longer be enough.The conversation also explores how AI is changing leadership, why five-year business strategies are becoming harder to defend, and what skills will matter as intelligent systems become more capable.

In this episode:

• Why human agency matters in the age of AI

• What happens when machines start making decisions for us

• How AI twins could help protect human autonomy

• Why AI’s impact matters more than its intelligence

• Whether losing control of AI is inevitable

• Who takes responsibility when autonomous systems make decisions

• The rise of superintelligence and intelligent infrastructure

• What skills will matter in an AI-driven world

• Why traditional business strategies need to change

• How leaders can prepare for the future of AI

Chapters:

00:00 Introduction: The rise of human agency

03:31 How do we preserve human autonomy?

05:08 The concept of an AI twin

06:34 Can we measure AI's impact on society?

07:48 Are we asking the wrong questions about AI?

10:58 Can we avoid losing control of AI?

13:21 Who is responsible for autonomous AI?

14:46 Superintelligence and the future of humanity

17:43 What skills will matter in an AI-driven world?

19:47 Knowledge vs knowing how to apply it

21:12 How businesses need to change

21:44 Why long-term strategy is becoming obsolete

23:28 Strategy inside intelligent infrastructure

25:17 What leaders should do in the next 12 months

26:00 Final thoughts

About BuildersBuilders is a podcast by Proxify exploring the people, ideas, and technologies shaping the future of business and technology. Hosted by Lee Haynes, each episode features conversations with founders, executives, and technology leaders about building, leading, and adapting in a changing world.

Subscribe for more conversations about AI, technology, leadership, and the future of business.

What if the future of AI in business isn’t about adding AI to your existing software, but replacing the software altogether?

In Part 1 of this conversation, our host Lee Haynes sits down with technology executive, board member, and author Ingo Paas to examine why most companies may be approaching AI transformation from the wrong direction.

Organizations are adding AI to existing processes, SaaS applications, and fragmented technology stacks. Ingo argues that this misses the bigger shift. If AI can take over the processes those applications were built to support, much of today’s SaaS stack could eventually disappear.They also discuss why traditional governance struggles to keep pace with AI, why boards are often not asking the right questions, and why three and five-year strategies become increasingly difficult when the underlying technology can change dramatically within months.

In this episode:

• Why traditional governance struggles with AI

• The growing transfer of decision-making authority to machines

• What boards are missing about AI transformation

• Why AI could make organizational fragmentation worse

• How leaders should balance experimentation with running the existing business

• Why adding AI to existing SaaS may be the wrong approach

• Ingo’s case for replacing SaaS applications with AI

• Why companies need to start redesigning from the core

Chapters:

00:00 Introduction

01:42 Ingo Paas’ background

03:06 Why AI transformation is only beginning

04:38 Why traditional governance fails with AI

06:45 What AI governance should look like

09:10 Governing AI in real time

11:28 The data problem companies already have

13:50 When machines start making decisions

14:40 What surveillance and warfare teach us about AI

19:10 What boards are getting wrong about AI

22:27 Why AI requires a different kind of leadership

24:32 The problem of fragmented organizations

25:54 Is AI the end of SaaS?

28:52 Experimentation vs paralysis

What does it take to build a career in AI when the technology changes faster than any university curriculum can keep up?

Joana Fonseca is an AI engineer at TRATON with a PhD in machine learning and robotics from KTH Royal Institute of Technology. Yet even after a bachelor’s, master’s, and PhD, she had never learned or used generative AI. When she finished her PhD, she had to learn transformers and generative AI herself. In this episode of Builders, Joana talks about the foundations AI engineers need, why continuous learning is unavoidable, and what separates someone who can build an impressive prototype from someone who can build reliable, valuable AI systems.

She also discusses the importance of domain collaboration and data engineering, what makes a company truly AI-native, where vision language models (VLMs) and vision language action models (VLAs) are heading, and what AI could look like by 2030.

Joana also shares what she is seeing through Stockholm AI and why the conversation around AI increasingly extends beyond models into infrastructure, governance, policy, and societal impact.

Chapters

00:00 VLMs and autonomous driving

00:34 Introduction

00:48 What does it take to become an AI engineer?

01:38 The foundations every AI engineer needs

02:03 Why continuous learning is unavoidable

02:37 Even PhDs had to learn generative AI

03:05 Learning AI after a PhD

04:39 How important is domain expertise?

05:04 Why collaboration matters in AI

06:15 What separates strong AI engineers?

06:31 Building AI beyond the prototype

07:04 The importance of data engineering

07:17 What the AI community is talking about

09:30 AI's impact beyond technology

10:50 What makes a company AI-native?

12:30 What's next for VLMs and VLAs?

13:30 The causality problem in autonomous driving

16:00 How researchers are improving VLAs

17:00 What will AI look like in 2030?

19:00 Advice for the next generation of AI engineers

19:22 Why lifelong learning matters

Joana Fonseca is an AI engineer at TRATON, working on AI for autonomous driving.

In this episode of Builders, Joana explains how vision language models (VLMs) and vision language action models (VLAs) could change autonomous driving by helping vehicles understand entire scenarios rather than simply detecting individual objects.

She also breaks down what makes deploying AI in the physical world so difficult, from latency and model size to safety, edge cases, testing, and the huge amounts of diverse real-world data required.

Joana holds a PhD in machine learning and robotics from KTH Royal Institute of Technology. Her work has taken her from developing algorithms for autonomous submarines to building AI systems for autonomous trucks at TRATON.In this episode:

• How VLMs differ from traditional computer vision
• How AI can understand an entire driving scenario
• The role of VLAs in autonomous vehicles
• Why 99.99% reliability may still not be enough
• Why latency becomes critical when AI operates in the physical world
• The data problem behind autonomous driving
• Why moving from prototype to production remains so difficult
• What production-ready AI looks like for autonomous systems

Subscribe to Builders for more conversations with the people building the future of technology, AI, and engineering.

Chapters

(00:00) Introduction

(01:41)⁠ From robotics to autonomous driving

(03:49)⁠ Building AI for autonomous trucks

(⁠04:47)⁠ What are vision language models?

(⁠05:48)⁠ VLMs vs traditional computer vision

(⁠07:34)⁠ From VLMs to vision language action models

(⁠09:51)⁠ The black box problem

(⁠12:38)⁠ Why 99.99% safety isn’t enough

(⁠13:57)⁠ When AI latency becomes dangerous

(⁠15:01)⁠ Testing AI in the physical world

(⁠16:12)⁠ The autonomous driving data problem

(⁠17:28)⁠ Where current AI models fall short

(⁠18:49)⁠ Why prototypes struggle to reach production

(⁠19:52)⁠ Why robotaxis haven’t scaled everywhere

(⁠21:36)⁠ What makes AI production-ready

AI is rapidly evolving from a powerful tool into a transformative force for organizations and society.In this keynote, Jesper Fredriksson explores the current state of AI, the accelerating pace of technological advancement, and the challenges organizations face when moving from experimentation to large-scale adoption.A central theme is the "last mile" problem—the gap between impressive AI capabilities and fully automated, real-world outcomes. Jesper discusses how AI is reshaping software engineering, productivity, and decision-making, while highlighting the importance of human oversight, guardrails, and responsible implementation.In this episode:- AI's evolving capabilities and future potential- The last mile challenge in automation- AI's impact on software engineering and productivity- Technical and organizational barriers to adoption- Human oversight and guardrails in AI systems- Risks and ethical considerations of advanced AI- Moving from AI experimentation to measurable outcomes#AI #Automation #ArtificialIntelligence #SoftwareEngineering #Productivity #FutureOfAI #AIInnovation

I agents are becoming more capable and more autonomous. But as they gain access to tools, memory, and company data, they also introduce new security risks.

In this Proxify Talk, Ali Leylani, Co-founder & Chief AI Officer at Echo Alpha and Chairman of Stockholm AI, explains how AI agent systems work, where they’re vulnerable, and what engineering teams should think about before deploying them at scale. Drawing on his background in AI, cybersecurity, and defense technology, Ali explores the emerging attack surface created by autonomous and multi-agent systems.

In this episode, you’ll learn:

• How AI agents combine language models, memory, and external tools to perform complex tasks
• Why multi-agent systems introduce new security challenges
• How prompt injection, jailbreaks, and other attack techniques target AI systems
• Why a single compromised agent or data source can affect an entire workflow
• How malicious instructions can be hidden inside images and other multimodal inputs

• What organizations should consider before giving AI agents greater autonomy.

Whether you’re building AI products, leading an engineering team, or exploring agentic workflows, this talk offers a practical look at one of the biggest challenges facing AI adoption today.

#AISecurity #ArtificialIntelligence #CyberSecurity #AIAgents #ProxifyTalks

AI implementation is easy. AI adoption is hard.In this episode, Louise Vanerell and Carl Carlheim-Gyllensköld explore why so many AI initiatives fail to create lasting organizational value despite strong technology investments. The conversation focuses on a critical but often overlooked element of successful AI transformation: the human layer.Together, they discuss how organizations can move beyond technical deployment and drive real behavioral change through effective communication, psychological safety, change management, and internal champions.Whether you’re leading an AI transformation, building an AI strategy, or helping teams adapt to new technologies, this episode offers practical insights for turning AI adoption into measurable business outcomes.In this episode:

• Why AI adoption often fails in organizations
• The importance of engineering the human layer
• Building trust and psychological safety around AI
• Effective communication and change management strategies
• The role of champions in technology diffusion
• Driving behavioral change across teams
• Moving from AI usage to organizational value
• Practical lessons for leaders navigating AI transformation

If you enjoyed this conversation, subscribe for more discussions on AI strategy, organizational transformation, leadership, and the future of work.#AIAdoption #ArtificialIntelligence #ChangeManagement #AIStrategy #DigitalTransformation #Leadership #FutureOfWork #Innovation #BusinessTransformation #AI