In this episode of The AI Moment, I sat down with Jonathan Wagstaffe to tackle one of the most pressing questions for modern business leaders: Is it time to build your own company LLM? We move past the hype of "building from scratch" to discuss the practical realities of the Rent, Buy, vs. Build framework.

We explore why context is the new currency in AI. It is no longer enough to simply use a public model; to gain a competitive edge, businesses need to integrate their own operating procedures and product ecosystem into the AI's workflow. However, this isn't without significant risk. We discuss the "dull, boring" but essential issue of data quality, noting that messy or fragmented data will undermine even the most sophisticated AI ambitions.

The conversation highlights Yahoo Scout as a leading example of the "hybrid model"—taking a powerful base like Claude and layering specific data on top to create a specialist tool. For leaders, the takeaway is clear: be mindful of the exploding costs of token usage and the scarcity of AI expertise. Instead of chasing a "naked LLM," focus on building the proprietary guardrails and intelligence layers that turn a generic tool into a powerful business asset. As the enterprise space evolves rapidly toward the summer of 2026, staying agile with a hybrid approach is your best bet to avoid being "cleaned out" by rapid platform shifts

In this episode of The AI Moment, Jonathan Wagstaffe and me (Danny Denhard) tackle the most pressing question facing modern executives: How do we actually measure the success of AI?

As businesses move past the initial excitement of generative tools, the challenge is to move away from "instinctive" utility and towards rigorous, actionable KPIs that satisfy the boardroom.

The discussion centres on moving the goalposts from measuring the AI itself to measuring the impact on existing business metrics. Danny introduces a robust four-pillar framework for leaders to adopt: Velocity, Quality, Economic, and Strategic. This includes looking at "Keep-Me-Out-Of-Jail" metrics like hallucination rates and the "Human-in-the-Loop" (HITOR) rate—measuring how much human intervention is required to make AI output viable.

>> We have go deeper in sharing more insights on our newsletter Subscribe to our newsletter at ⁠https://aimomentpodcast.substack.com/subscribe⁠ for deeper dives and intelligence delivered with every podcast episode.

Jonathan and Danny also explore the departmental nuances of these KPIs, noting that success in Sales looks very different from success in Support or Operations. Whether it is reducing proposal turnaround time or decreasing product decay rates, the message is clear: AI is a lever for business outcomes, not an outcome in itself.

Key Takeaways for This Week:

• Identify the "business results" you want before deploying the tool.

• Track "saved time" through the lens of what that time is reinvested into.

• • Establish "Trust and Reliability" metrics to manage hallucination risks.

• Enjoyed the episode? Subscribe and follow the podcast on your favourite platform to never miss an insight.Want more?

  
  

• 
  

In this mailbox episode, Jonathan (Wagstaffe) and I dive into a challenge many of you are facing: a CEO who is obsessed with AI but lacks a fundamental understanding of its limitations.

It is a classic case of managing up. When your leader is influenced by the "best-case" scenarios shared on social media, your job is to redirect that energy into grounded, actionable strategy.

We discuss the necessity of executive education, specifically through dedicated training days and hands-on hackathons.

These sessions are designed to pull back the curtain on the "magic" and show just how frustrating and difficult AI implementation can be when you move past the "vibe marketing" stage.

We also explore the dangers of "AI washing," where companies use the technology as a shield for headcount reductions, often leading to the eventual need to re-hire the talent they prematurely let go.

The episode provides a roadmap for shifting the conversation from "replacing jobs" to "replacing tasks" through workflow mapping and measured experiments. We wrap up by re-introducing the "3 Ts (Time, Truth, Trust) and 2 Vs (Validate and Verify) " framework for maintaining quality and reliability in an era of AI hallucinations. This is about ensuring your business moves at pace without sacrificing the truth

Thanks for listening!

Danny Denhard