Welcome to our end-of-year episode of Future Bytes with co-host Jade Emmanuel, Senior Data & AI Consultant at Columbus UK.

In this episode, we'll revisit insights from our most popular episodes this year. We'll share predictions for AI in business and wrap up by answering questions from listeners.

Thank you for following along. We can't wait to continue the conversation in 2026.

Happy New Year from all of us!

Hosted on Acast. See acast.com/privacy for more information.

In the final AI News by Future Bytes episode of the year, Magnus Oxenwaldt highlights what no longer moves the market. New model releases are landing quietly. Performance gaps are narrowing. The centre of gravity in enterprise AI is shifting.

Top stories for week 52:

• Model releases lose impact: Google’s Gemini 3 Flash launched with little reaction, reflecting how GPT, Gemini and Claude now sit within statistical distance for most business use cases.
• Workflows become portable: Anthropic opened “skills” as an open standard, allowing AI workflows to move across models and tools.
• Cloud players diversify exposure: Amazon is reportedly in talks to invest in OpenAI while already backing Anthropic.
• Platforms converge: Shared protocols, app marketplaces and autonomous research agents signal a more interoperable enterprise AI stack.

See you next year!

Hosted on Acast. See acast.com/privacy for more information.

AI is approaching an inflection point. Magnus Oxenwaldt speaks with Columbus CEO Søren Krogh about what this moment means for leadership. Which experiences shaped Søren’s approach to technology and leadership?

This conversation reflects a thoughtful approach to leadership in the AI era. Not about having all the answers, but about sound judgement. Knowing when to go deep, when to trust others, and how to stay focused amid the noise. The emphasis is practical and human. Real value over impressive demos. Empathy and responsibility as steady guides in a time of rapid change.

“For me, AI is possibilities," Krogh says.

Key themes:

• AI as a leadership question, not a technology project
• Turning possibilities into prioritised action
• What responsible, modern AI leadership looks like inside organisations

Hosted on Acast. See acast.com/privacy for more information.

This week’s AI News by Future Bytes covers two stories that reveal where enterprise AI is heading. Major AI providers formed a foundation to set standards for agent communication. OpenAI released GPT-5.2 after an internal “code red”. The focus is shifting from models to orchestration.

Top stories this week - cooperation vs. competition:

• OpenAI, Microsoft, Google, Amazon and others formed the Agentic AI Foundation to create open standards for how AI agents communicate.
• OpenAI released GPT-5.2, fast-tracked after Google Gemini and Anthropic gained ground, highlighting the limits of benchmark comparisons.

Hosted on Acast. See acast.com/privacy for more information.

Hosted on Acast. See acast.com/privacy for more information.

OpenAI issued a real "code red" as Google and Anthropic closed in. Gemini’s surge and Claude’s gains tightened the race, signalling a maturing market and rising pressure on enterprises.

– "Competition is working. Prices are falling. Capabilities are converging. For buyers, that’s the best news possible,” host Magnus Oxenwaldt says,

Top stories for week 50:

·  OpenAI declared code red as Google and Anthropic closed the gap, forcing Sam Altman to pause multiple upcoming projects.

·  Google launched Gemini 3 Pro and its new DeepThink reasoning tier, prompting Geoffrey Hinton to say Google may now overtake OpenAI.

·  Anthropic struck a $200M Snowflake partnership and is exploring an IPO, positioning itself as the enterprise AI stack.

·  New data shows the enterprise deployment gap widening, with Microsoft missing internal AI sales targets and only 5% of AI projects scaling beyond pilots.

·  With GPT, Gemini and Claude converging on benchmarks, enterprise value now hinges more on integration with existing data infrastructure than raw model performance.

Hosted on Acast. See acast.com/privacy for more information.

Three major AI models launched this week: Google’s Gemini 3, Anthropic’s Claude Opus 4.5, and OpenAI’s GPT-5.1 Codex Max. Each lead in a different area, and the gap between benchmarks and real-world performance is widening. 

The bigger shift? AI is moving into specialisation across the entire stack. Microsoft is betting on being the orchestration hub. Google is leaning on vertical integration. And the market is fragmenting faster than expected. 

Top stories this week:

• Google, Anthropic and OpenAI all launched new models this week, each claiming leadership in a different area. 
• Benchmarks diverge from real-world experience — developers still prefer certain models for how they reason and solve edge cases. 
• AI is shifting toward specialisation: voice, image, coding, and text handled by different best-in-class services. 
• Microsoft positions itself as the orchestration hub with Foundry and its investments in Anthropic and OpenAI. 
• Google’s vertical stack and potential sale of TPU chips to Meta show a market that is fragmenting, not consolidating. 

Hosted on Acast. See acast.com/privacy for more information.

This week, the EU postponed its core AI rules by sixteen months, pushing them into 2027. But while regulators slowed down, technology accelerated. At the same time, Microsoft and Google introduced tools that let anyone build autonomous agents, automate workflows, and create apps through simple conversation. Hyper-automation moved into the mainstream. Regulation fell even further behind. And enterprises now face a widening governance gap they will have to close themselves.

Top stories this week:

1. EU delays high-risk AI rules to 2027–2028, leaving a long regulatory gap.
2. Microsoft and Google push no-code agents mainstream, letting anyone build automations.
3. Hyper-automation accelerates, with agents now able to write and validate their own code.
4. Enterprise adoption spikes: 99% of developers build agents; 40% of big firms deploy them this year.
5. Governance falls behind, forcing companies to build their own internal AI controls.

Hosted on Acast. See acast.com/privacy for more information.

This week, something unexpected happened. Warren Buffett, the investor who avoided tech hype for 20 years, just bought $4.9 billion worth of Alphabet shares. 

At the same time: 

• Analysts warn the AI economy may already resemble the dot-com bubble 
• Nvidia is hitting record highs while AI infrastructure spending explodes 
• Most enterprises still report no measurable ROI from AI 
• Sub-prime winner Michael Burry is betting against AI infrastructure companies 

So, what’s real? A bubble? A breakthrough? 

Or both at the same time? 

Hosted on Acast. See acast.com/privacy for more information.

After years of intense startup work and long-haul flights, a doctor told Lars Tvede to take a year off. He moved to the Swiss mountains to rest. Instead, he ended up starting two new companies. He simply couldn't help it. It is what makes him happy.

That instinct hasn't changed. Today, Tvede is behind Supertrends, a platform that uses AI to help businesses make better predictions about the future. In this episode, host Magnus Oxenwaldt dives into the philosophy behind Supertrends. We also talk about reinventing the media with an AI-powered news and insights platform built around "breaking views" rather than breaking news.

If you want to understand where technology is heading, this conversation is worth listening to.

Key ingsights:

·     Supertrends monitors 5,000 publications daily in 40 languages, tracking 4,000 technology predictions

·     The AI system now handles work that previously required 160 human experts

·     China remains the biggest challenge – Tvede calls it his greatest blind spot for emerging tech

·     New venture “Project Y” aims to launch an AI-powered media company in Denmark

Links mentioned in the podcast:

www.supertrends.com

www.projecty.dk

Hosted on Acast. See acast.com/privacy for more information.

AI in business is moving from the cloud to the real world: With humanoid robots entering the market at enterprise-software price points, robotics is no longer just for factories. Magnus Oxenwaldt explores what that means for 2027 budget planning: from SAP’s integration of robots into enterprise systems to the rise of remote robot operators in retail.  

This week's top stories for AI in business:  

• 1X Technologies opens pre-orders for NEO, a $20,000 humanoid robot. A price that makes robotics accessible beyond manufacturing.  
• SAP integrates humanoid robots directly into its warehouse management system, signalling physical automation’s entry into enterprise software.  
• Figure AI robots now run full shifts at BMW plants, while Amazon hits one million deployed robots across 300 sites.  
• Microsoft warns of GPU power shortages as physical AI multiplies cloud demand, and Google plans orbital AI data centres by 2027.  
• The strategic takeaway: digital transformation now has two tracks, software automation and physical automation. It’s time to plan for both.  

Hosted on Acast. See acast.com/privacy for more information.

Something shifted this week. OpenAI’s Sam Altman revealed their internal target: a fully automated AI researcher by March 2028. That’s 29 months away, and big tech is already moving as if it’s real. 

This week’s takeaways: 

• Tech giants are reallocating capital for a 2028 AGI milestone 

• Infrastructure bets: 10–30 GW in AI compute by OpenAI & NVIDIA 

• Job cuts signal a pivot from people to machine-driven R&D 

• OpenAI planning IPO ahead of AGI moment — valuation timing matters 

• Businesses should model two AI futures: slow vs accelerated 

Hosted on Acast. See acast.com/privacy for more information.

This week, the AI industry received a major dose of realism from OpenAI co-founder Andrej Karpathy. Speaking on the race towards artificial general intelligence (AGI), Karpathy said we’re still a decade away from seeing it work as intended, calling today’s AI agents ‘slop’. For businesses that have held back on their AI investments, fearing that AGI might soon render their efforts obsolete, Karpathy’s comments flip that logic on its head. If AGI is ten years out, the excuse for waiting disappears.

Top stories this week:

·       Claude adds long-term memory, enabling models to recall user projects and preferences across sessions.

·       Google’s Willow chip achieves quantum advantage, running algorithms 13,000× faster than the top supercomputer.

·       Anthropic signs a $40 billion TPU deal with Google, securing compute capacity for the AI decade.

·       Oracle launches an AI agent marketplace, while Microsoft opens an Agent Store.

·       DeepSeek unveils OCR models compressing context windows by 10×, and GPT-5 sets new SWE-Bench and AIME benchmarks.

Hosted on Acast. See acast.com/privacy for more information.

This week, enterprise AI stopped being optional. AI is now part of the system, not a tool you choose. “October 15th wasn’t the day AI got better. It was the day the industry removed every excuse you had for not adopting it,” host Magnus Oxenwaldt says. In this week’s AI News by Future Bytes, he breaks down how these moves are reshaping business, productivity and competition. 

Top stories this week:

• AI just became the default. It’s now built into Windows, Microsoft 365, and the tools you already use.
• The excuses are gone. Infrastructure, cost, and integration barriers disappeared overnight.
• Free isn’t free. Bundled AI means platform lock-in and long-term dependence.
• The game moved fast. Early adopters are already compounding their advantage while others are still “exploring use cases.”

Questions or feedback? Reach out to julie.malvik@columbusglobal.com 

Hosted on Acast. See acast.com/privacy for more information.

China just became the world’s largest AI market. While the US bans Chinese tools, researchers still collaborate. Europe talks sovereignty but relies on both sides.

“Fear just became bigger than geopolitics. The AI race isn’t about who wins. It’s about whether humanity can build intelligence safely."

Top stories this week:

• China overtakes the US with 195 million AI users

• DeepSeek builds powerful models on restricted chips

• US bans Chinese AI tools, yet research ties remain

• Europe’s AI sovereignty meets economic reality

• OpenAI, Anthropic and Microsoft tighten platform control

Got questions or want to share your thoughts? Reach out to producer julie.malvik@columbusglobal.com or directly to Magnus Oxenwaldt at magnus.oxenwaldt@columbus.com.

Hosted on Acast. See acast.com/privacy for more information.

This week on AI news by Future Bytes, host Magnus Oxenwaldt breaks down how OpenAI’s latest moves, Sora 2 and Instant Checkout, signal a shift from apps to AI platforms. TikTok, Google, and even your own website could be next. Welcome to the platform trap.

Top stories:

• Sora 2’s launch turns AI-generated video into social media, reaching the US App Store’s top spot overnight. 
• ChatGPT adds Instant Checkout, enabling shopping, payments, and product search, all without leaving the app. 
• The rise of the “platform trap”. Every major AI company is building its own closed ecosystem for work, commerce, and creativity. 
• Meta’s Business AI connects 700 million consumers with always-on sales agents across all Meta platforms. 
• Microsoft’s Agent Mode boosts Copilot accuracy and brings fully autonomous AI agents into everyday productivity tools. 
• Salesforce’s Agentforce aims for one billion agents by 2025, extending AI into every customer touchpoint. 
• Google fights to protect $273 billion in ad revenue as users turn to AI assistants instead of search. 
• The open web is at risk. Businesses that fail to integrate with AI ecosystems could disappear from the customer journey entirely. 

Hosted on Acast. See acast.com/privacy for more information.

Stay ahead. Each week, host Magnus Oxenwaldt, breaks down the top AI stories and what they mean for your business. 

Top stories this week: 

• Nvidia + OpenAI $100B: Partnership to build 10GW of AI data centres for the $500B “Stargate Project” – a scale bigger than some nations’ power use. 
• CoreWeave hits $22.4B: Specialist cloud wins with fastest GPU access. 
• Oracle raises $15B: Debt fuels the AI infrastructure race. 
• 90% using AI: Tech staff adopt tools daily, often without approval. 
• AI’s double edge: Cancer detection breakthrough – but also synthetic virus design. UN moves on governance. 

Hosted on Acast. See acast.com/privacy for more information.

AI moves fast! But don’t worry, we’ve got you covered. AI News by Future Bytes brings you the most important AI news for business in just 10 minutes a week. The biggest shifts, the smartest moves, and the practical takeaways you need to stay ahead. Hosted by Magnus Oxenwaldt, AI Director at Columbus.

Top stories this week:

• Oracle bets $300B on AI compute: transforming from laggard to infrastructure powerhouse.
• Microsoft adds Anthropic’s Claude to 365 Copilot: giving enterprises new options and a multimodal moat.
• Reproducible AI breakthrough from Thinking Machine Labs: making AI predictable enough for mission-critical use.
• Apple & Meta hardware race: real-time translation AirPods and AR glasses signal a new era of AI interaction.

Hosted on Acast. See acast.com/privacy for more information.

Any transformation takes courage, whatever the technology. “It’s about reimagining possibilities, using data and imagination to drive change, and showing every stakeholder what’s in it for them,” says Ian Kingstone, Business Transformation Advisor at Columbus.

“When people align early on what ‘good’ looks like and can see their role in achieving it, they start pulling the change rather than having it pushed onto them,” he says.

In transformations involving AI, management often chases cost efficiencies. Kingstone says those will come — but if you really want a competitive edge, you need to be bold and growth-oriented.

“The organisations that win won’t just copy; they’ll imagine new possibilities. That requires aligned mindsets and a clear vision. AI can be abstract and hard to visualise, so you need experimentation, testing, and learning. Be brave!”

Reach out to Ian Kingstone on LinkedIn at: linkedin.com/in/iankingstone.

Got questions or want to share your thoughts? Reach out to producer julie.malvik@columbusglobal.com or directly to Magnus Oxenwaldt at magnus.oxenwaldt@columbus.com.

Hosted on Acast. See acast.com/privacy for more information.

AI isn’t here to replace you. It’s here to work with you. In this episode, Magnus Oxenwaldt looks at what it means to have AI as part of the team. From mindset shifts to structural change: what needs to change in how we think, work, and organise? Tune in for practical insights on welcoming your next digital teammate.

Got questions or want to share your thoughts? Reach out to producer julie.malvik@columbusglobal.com or directly to Magnus Oxenwaldt at magnus.oxenwaldt@columbus.com.

Hosted on Acast. See acast.com/privacy for more information.