Show Notes: The 2025 AI Year in Review

Jonathan and I are sitting down to wrap up what has been a truly transformative year for the AI landscape

If 2024 was the year of hype, 2025 was the year of the "Great Separation," where we finally saw which players could execute and which were just blowing smoke.

This episode is a longer pod so here are the chapters to jump to if you are pushed for time:

00:00 Introduction and Year-End Reflection

00:22 OpenAI's Mixed Year

04:35 AI Market Trends and Investments

06:47 Google Gemini's Progress

09:48 AI Skillset and Workforce Impact

12:05 Microsoft Copilot's Struggles

16:21 Power and Infrastructure Challenges

19:38 Apple and Meta's AI Journey

22:45 AI Application Success Stories

28:52 Conclusion and Future Outlook

The Big Shifts of 2025

We dive deep into why Google Gemini has arguably won the year. While they started weak, they’ve successfully re-engineered their entire organisation around an AI ecosystem, delivering incredible tools like NotebookLM and Nano Banana.

On the flip side, OpenAI has had a mixed scorecard.

Despite holding 70% market share, they’ve faced a "Code Red" at year-end, pivoting away from broad "empire mode" experiments to protect their core LLM.

We also tackle the "Microsoft Problem".

Despite its reach, Copilot is struggling with user delight due to heavy corporate guardrails, leading to a disappointing adoption rate compared to sleeker apps like Gamma.

Key Takeaways

• The Skillset Advantage: Only 10-12% of knowledge workers are truly using AI. If you are investing time to learn these tools now, you are already in the global elite.
  
• Infrastructure is the New Wall: The West is facing a massive power generation crisis
• OpenAI alone may soon require energy equivalent to one-fifth of the US's current output.
  
• Time to Value (TTV): Tools like Gamma are setting the standard for "Time to Edit," allowing you to move from a recorded conversation to a finished strategy deck in minutes.
  

1. Stop the Start-Stop: If your company is stalling AI rollouts for "legal sign-off," you are losing ground every day.
   
2. Try NotebookLM: It is the best free tool available for turning your messy data into actionable insights.
   
3. Subscribe: Join our community at aimoment.co.uk for deeper dives every Monday and Friday.
   

In this episode of The AI Moment, Jonathan Wagstaffe and I dive into a common failure mode we are seeing across the business landscape: the ‘start-stop’ motion of AI adoption.

Too many businesses treat AI implementation like a pit stop—they pull in to perfect their strategy or policy, but then struggle to get back into the race.

We discuss the real danger of ‘over-perfecting’ governance. While some leaders pause to hit 100% policy perfection, competitors and internal outliers race ahead at 70% readiness.

This creates a massive internal disconnect; when the company hits the brakes, enthusiastic employees often just go underground, building their own ‘shadow AI’ toolkits while the rest of the business stagnates.

Key Takeaways:

Measure Twice, Cut Once: Jonathan emphasises the importance of the ‘thinking phase’ before experimentation begins. Spending time defining playbooks and guidelines upfront prevents the frustrating scenario where Legal steps in weeks later to ban work that has already started.

Adopt 90-Day Sprints: To cure the start-stop cycle, we recommend working in 90-day sprints. Set your scope, run the sprint, and evaluate at the end. This structure prevents the distraction of constantly debating tool switches: like the recent Gemini vs. ChatGPT discourse & keeps the team focused on execution.

Leadership Clarity: Leaders must define three core themes and stick to them. This stability allows your AI champions to guide the rest of the team to maturity without the agenda constantly shifting.

Your Call to Action!

Stop aiming for a perfect policy that doesn’t exist. Establish your 90-day goals, empower your champions, and keep the momentum going.

If you enjoyed this discussion, please rate and review us in your podcast player.

For deeper analysis on these topics, Please subscribe to our supporting newsletter at aimoment.co.uk.

Thanks for listening

Danny Denhard

2026 Predictions: The Big Hack, Agentic Shopping, and Multiplayer AI

As we close out 2025, Jonathan and I decided it was time to get the crystal ball out and predict where the world of AI and business is heading in 2026.

While 2025 was about experimentation, 2026 is going to be the year of ROI, integration, and some serious security challenges.

Key Takeaways:

• The Big AI Hack is Coming: My first prediction is a sombre one. We have already seen the Claude MCP hack, but I believe we will see a major, targeted attack on the big LLMs or agentic systems in 2026 . Hackers follow the money, and with agents now able to take action on our behalf, the "black hat" incentive is huge .

• The End of "Blue Links": Jonathan predicts a massive shift in e-commerce. We are moving away from searching Google for a list of links to asking an AI to "find the best lawnmower" and having it present a single buy button . This means brands must offer pricing consistency; if your prices jump around, AI agents simply won't recommend you .

• Multiplayer AI: I believe the days of "single player" AI are numbered. 2026 will be the era of "Multiplayer AI," where teams collaborate with LLMs inside shared workflows rather than isolated chat windows . This will finally drive better cross-functional connection .

• Democracy’s Acid Test: We also discussed the upcoming US midterms . With the cost of creating deepfakes and AI content dropping to near zero, we expect a flood of disinformation that will test the limits of voter trust .

Take Action:

1. Audit Your Security: If you are using AI to write emails or manage tasks, double-check every hyperlink . Security against AI-assisted hacking must be a priority for 2026 .

2. Prepare for Agents: If you work in e-commerce, start thinking about how your product data appears to a machine, not just a human .

   
   

   Join the Conversation: We’d love to hear your own predictions. Drop us a line at ai@dannydenhard.com and don't forget to subscribe to our supporting newsletter at aimoment.co.uk for deeper analysis .

• 
  

• We will see you next time!