Welcome to The NeuralPod. The 0-1 machine learning podcast. Chatting with ML Leaders, Researchers and Engineers who've built models, systems and products 0-1.
What does sovereign AI really mean for the UK? Adie Holt of Cosine joins The NeuralPod to talk UK compute, Lumen Sovereign, and why agency over your AI models matters.
Adie spent more than two decades as an RAF engineer before joining Sir Chris Deverell's jHub innovation unit in 2017. He shares what that taught him about how large institutions adopt new technology, why leaders leaving the military should aim for confident humility, and how corporate short-termism holds back AI adoption.
We then get into sovereign AI: freedom of action, access and control over the models you deploy, and why relying only on trusted access to US models is becoming riskier for the UK, its NATO partners and regulated industries. Adie explains how Cosine went from a not for profit AI lab to code engineering for banking, defence and security, training models on legacy languages like COBOL, Fortran and Ada. He also covers Lumen Sovereign, the British large language model Cosine is training on the Isambard AI supercomputer in Bristol with full data provenance, and Sparkler, the post-training tool the team built under tight compute constraints. Plus the UK compute gap, distributed data centres, and what government should do now to keep Britain in the AI race.
Chapters
0:00 The UK's AI choice: customer or contender
1:06 Welcome Adie Holt of Cosine
1:57 From RAF technician to jHub innovation
3:50 Military vs corporate technology adoption
6:34 Leadership lessons: confident humility
9:20 What sovereign AI really means
13:09 Why nations want their own sovereign AI model
14:59 Can the UK compete with US AI spending?
18:03 Isambard AI and greener AI compute
19:12 Cosine's story: from AI lab to regulated industry
21:43 Trust, data provenance and AI safety
23:57 Sparkler: training models on limited compute
26:09 Lumen Sovereign: a British general purpose LLM
28:04 Why choose Cosine: agency and control
29:33 Predictions for UK AI
32:51 What the UK should do now on compute
34:36 Book recommendations
Quotes
"The government has given the UK a dog in the AI fight."
"Constraints can be a great engine for innovation."
"If we back artificial intelligence today, we put ourselves on the top table of one of the most important and defining technologies of the next 100 years."
Books mentioned
So Good They Can't Ignore You by Cal Newport
Incorruptible: Why Good Companies Go Bad and How Great Companies Stay Great by Eric Ries
Connect:
Adie Holt on LinkedIn: https://www.linkedin.com/in/adrianflholt/ (https://www.linkedin.com/in/adrianflholt/)
Cosine: http://linkedin.com/company/91573134/ (http://linkedin.com/company/91573134/)
Chris: https://www.linkedin.com/in/chris-ml (https://www.linkedin.com/in/chris-ml)
Subscribe to The NeuralPod for more conversations on how real machine learning systems get built.
#SovereignAI #UKAI #ArtificialIntelligence
7 May 2026
Darminder Singh - Scaling Agents and AI from Pilot to Production
How do you actually move enterprise AI from pilot to production — and measure real ROI?
In this episode of The NeuralPod, Chris Coyne sits down with Darminder Singh, AI leader at HCLTech AI Labs, to unpack what separates AI initiatives that deliver measurable business value from those that stall.
With a career spanning engineering, data science at KPMG and Fujitsu, and deep work in defence and critical national infrastructure, Darminder offers a rare cross-sector view on what genuinely works.
The conversation covers:
00:00 Why Leverage AI
00:23 Meet Amin Inda
00:54 Engineering to Data Value
02:55 Business Skills for AI Leaders
04:31 Consulting Lessons and Teams
05:36 Hiring for Growth Mindset
08:03 AI in Defense Edge vs Office
10:34 Data Culture and Tool Ecosystems
15:11 HCL Labs and Scaling GenAI
17:45 Agent ROI and Orchestration
23:31 Data Mesh and Common Pitfalls
26:24 Guardrails Red Teaming Regulation
28:19 Future Trends Personal Agents
31:38 Autonomous Employees and Trust
32:47 Reading List and Wrap Up
11 Mar 2026
Vivek Chand - Clawbot, Agents and Observability
AI Agent Observability: OpenClaw, ClawMetry and Observability | Vivek Chand (Booking.com)
How do you monitor AI agents in production?
In this episode, Vivek Chand, tech leader at Booking.com and founder of ClawMetry breaks down AI agent observability, evaluation loops, and his open-source tool ClawMetry tracking agent behaviour in real time. Clawmetry is an open-source tool with already 30,000 plus downloads after just a week.
Vivek shares his engineering journey from startups to leading customer support AI at Booking.com, what it actually takes to ship AI systems that deliver ROI, and why observability is the missing layer for agentic AI workflows.
We cover OpenClaw for coding agents, how ClawMetry visualizes agent actions, subagent spawning, file access, and token usage, plus the roadmap for remote monitoring, mobile apps, and security alerts.
🔑 Topics covered:
AI agent observability explained
Building production AI systems at Booking.com
Evaluation loops for prediction quality
OpenClaw use cases for coding agents
Clawmetry Telemetry: open-source agent monitoring
Advice for junior AI engineers in 2026
00:00 Welcome and Guest Intro
00:59 Vivek Career Journey
03:03 Building AI at Booking
04:26 Scaling Systems and Eval Loops
06:19 Advice for Junior Engineers
07:48 Why Agent Observability Matters
11:30 OpenClaw Use Cases for Coding
13:21 Productivity Gains with Agents
14:43 What Telemetry Does
18:03 How Telemetry Was Built
19:53 Roadmap Managed and Security
22:24 Future of Agents and Monitoring
Reach and audience
Public platform figures. Ratings count people who left a rating, not total listeners.
YouTube views
3,165
Score snapshot 15 Sept 2025
Spotify
5.0 / 5
2 ratings
Podcast Authority Score: 34 / 100
A composite of feed quality, social presence, YouTube performance and engagement. Read the methodology.
Quality
41
Social presence
0
YouTube
68
Engagement
0
Contact 'What AI Means for Us'
Guest appearances
Does not typically book guests
Based on episode analysis; this does not confirm that the show is currently accepting guests.
Host of 'What AI Means for Us'?
Claim your podcast to manage its listing and keep your show details accurate.
Pod Engine is an independent podcast discovery and analytics service and is not affiliated with or endorsed by this podcast. Artwork and show content belong to their owners. Full legal notice.
Explore this show Podcast research with Pod Engine
