Hosts: Srini Annamaraju & David Royle

“Evals are the weak link in enterprise AI adoption.”

And we say it like it is in our Maven cohort Lightning Lesson. Enrol here or see the recording - or join the waitlist for the paid 4-part course (tba):

https://shorturl.at/lA9ig

This episode is a proper grilling on AI Evals: what they are, why boards should care, and why “ship it now, eval it later” is how you end up with a quiet disaster.

We also do a quick sweep on vendors going more “enterprise-native” (less benchmark theatre, more workflow reality).

What we cover

• Enterprise AI news: vendors shifting from benchmarks to enterprise workflows
• OpenAI’s Enterprise report highlights
• UiPath as the “plug-in hybrid” of automation: deterministic RPA meets GenAI via connectors (and why that blend might win)
• What evals actually are: accuracy, citations, groundedness, hallucinations
• Vendor reality: some push AI first and worry about evals later, others oversell eval tooling. Error analysis still matters
• Evals as the connective tissue between value, risk, and operations. Proactive, not post-mortem-after-the-horses-bolted
• The EDSO “four hats” operating model (Echo, Delta, Sigma, Omega) and why boards need the Omega translation layer
• Maturity and scaling: small firms can fuse hats, even one-person pods for bounded scopes
• Agentic future: “checker agents”, Delta agents writing eval harnesses, humans steering fleets of agents
• Why SMEs lag, and how eval expectations will percolate through supply chains

Chapters

• 00:02 Intro: Episode 7, cold UK afternoon, messy middle of enterprise AI
• 00:56 AI news: enterprise context is the new battleground
• 02:45 OpenAI Enterprise report headlines
• 10:16 UiPath, hybrid automation, and the “plug-in hybrid” analogy
• 12:53 The grilling starts: what are evals?
• 17:02 Is AI risk being exaggerated to sell governance tools?
• 19:45 Evals as connective tissue, and why proactive matters
• 21:55 The EDSO roles and what “good” looks like
• 25:21 Maturity levels and how smaller firms scope it
• 26:58 Checker agents and agentic operating models
• 28:58 Business case problem: cost vs avoided disaster
• 32:14 Evals in SMEs and supply-chain pressure
• 33:26 Close: “survived the grilling”

Takeaways

• Evals are not paperwork. They’re how you keep the value chain connected to operations without risk blowing up later.
• Don’t let vendors sell you “tooling-as-a-substitute-for-thinking.” You still need human error analysis and clear accountability.
• Treat EDSO as hats, not headcount. Start bounded, prove value, then scale.
• Evals is becoming a career lane (think “AI eval controller” the way finance has controllers).
• The agentic world will add “checker agents” and automated harness-writing, but humans still steer the system.

Who it’s for

CIOs, CDOs, CAIOs, Heads of Risk, and anyone trying to ship enterprise AI without quietly lighting their control environment on fire. Also, anyone building a real career edge around AI trust and operational quality.

Hosts: Srini Annamaraju & David Royle.

“The AI bubble is the wrong fear.”

The real threat sits inside your own walls: shadow AI you don’t see, boards that confuse risk aversion with risk management, and leaders trying to govern a technology they don’t actually understand.

We unpack why mid-market boards are exposed, how shadow AI reveals the truth about how your org really works, and what an actually realistic 12-month AI plan looks like.

And yes—why people, not models, are now the biggest AI risk vector.

The conversation revolves around a recent paper that David authored, a link to the post that has the details is here.

What we cover

1. Bubble noise vs fundamentals - Valuations swing wildly, but enterprise AI maturity rises daily. We explain why it has nothing to do with the technology reshaping your org.
2. Shadow AI as diagnosis - It’s not a tooling problem but a symptom of mismatched expectations.
3. Boards: from passive listeners to owners - Why literacy is step zero, and why chairs need to move fast.
4. Risk aversion trap - The boards that “get it” flip from “should we?” to “how quickly, safely, and visibly can we?”
5. 90-day governance playbook - Inventory → Validate → Govern.
6. Top-down vs bottom-up AI - How grassroots use cases and board-led operating models collide.
7. 12-month reality check - You won’t be AI-first in a year. But you can be an AI-literate, AI-safe, AI-enabled organisation in 12 months.
8. Explainability anxiety - Why boards demand transparency from AI they never asked of spreadsheets or humans.
9. The uncomfortable truth - The biggest AI risk isn’t the model. It’s your people.
10. Evals preview - Why audits, trust contracts, drift checks, and forward-deployed evaluators will soon be board-level concerns.

Chapters

• AI bubble vs enterprise fundamentals
• Shadow AI as a symptom
• Boards falling behind
• Risk aversion vs risk management
• 90-day governance plan
• A realistic 12-month AI horizon
• The real AI risk: people
• Intro to enterprise evals

Takeaways

• Shadow AI is a mirror - reveals gaps in culture, process, and leadership direction, not tooling.
• Boards must lead, not observe - Active literacy and ownership are key.
• Governance is the stabiliser. Inventories, validations, guardrails, and oversight reduce drift & exposure.
• Explainability is contextual. Set boundaries, not magic expectations.
• People are the attack surface. Don't miss non-malicious misuse.
• 12 months = foundations. Literacy, safety, and one high-value use case per function. That’s the win.

Who it’s for

Board members, CEOs, COOs, CIOs, CROs, and mid-market operators needing a grounded, real-world view of AI risk, governance, and organisational maturity.

Help Spread the Word - Enjoyed the episode? Follow the show, leave a review, and share with a colleague grappling with shadow AI, governance gaps, or board-level AI decisions. Want to join as a guest or sponsor a future episode? Get in touch!

Hosts: Srini Annamaraju & David Royle

“AI kills jobs” is the wrong headline. The real story is structural: org pyramids flatten into diamonds, managers run fleets of agents, SMEs unlock backlogs without hiring sprees, and skills go modular with micro-credentials.

We break down what changes now—and how to lead it without face-planting.

What we cover

• Jobs vs. roles: Why the entry-level layer thins, the manager layer thickens, and how to redesign spans of control when agents do the doing.
• Agents on a spectrum: Start with human-in-the-loop, graduate to AgentOps. Where to set autonomy today, what to monitor, and how to keep audits, drift checks, and safety rails sane.
• Backlog > headcount: Use AI to attack the work you never had people for—deterministic, high-volume tasks that finally move the needle.
• Operational resilience: Outages and dependency chains aren’t hypotheticals. We outline layered BCP/DR for an agentic stack so one failure doesn’t cascade.
• Early-career paradox: Apprenticeships still matter—how to select, coach, and rotate juniors in a world with fewer traditional entry roles.
• Skills that rise: Cognitive prompting, judgment, people leadership—and why short, role-tied micro-credentials beat semester-long generalities.
• SME timing & tactics: Where mid-market buyers actually are on the curve, what to build vs. buy, and how to avoid “pilot purgatory.”

Chapters

1. Jobs headline vs ground truth
2. From pyramid to diamond orgs
3. Agents, autonomy, and HITL → AgentOps
4. Managing hybrid teams (humans + agents)
5. Resilience playbook for outages and dependencies
6. Early-career design: apprenticeships, reverse mentoring
7. Micro-credentials and fast upskilling
8. What SMEs should do this quarter

Takeaways

• Jobs aren’t vanishing; roles are morphing. Plan for fewer juniors, more AI-enabled managers, explicit oversight of agent fleets.
• Governance is the unlock. Treat agents like teammates with performance records, audits, and clear escalation paths.
• Resilience is strategy. Design for failure before agents touch critical workflows.
• Upskill in sprints. Tie micro-credentials to roles, not buzzwords.

Who it’s for
Operators, CTOs/CIOs, and line leaders who need practical steps to re-shape teams, govern agentic workflows, and build real resilience—especially in SMEs.

Help Spread the Word:
Enjoyed the episode? Follow the show, leave a quick review, and share with a colleague wrestling with agent governance or workforce design. Interested in joining as a guest or sponsoring a future episode? Get in touch.

In this new Enterprise AI Field Notes deep dive, Srini Annamaraju (aka, 'the tech guy') and David Royle (who's 'the business guy') take the story past design into delivery — from the Target Operating Model (TOM) to the everyday reality of running AI inside the enterprise.

Through the lens of a real Bank's AI copilot rollout, name changed to "Albion Bank", they map how real transformation happens inside the Enterprise AI Honeycomb — a connected system of data, models, patterns, platforms, and guardrails that must all work in harmony.

💡 What we cover:

• Why do most AI “pilots” stall before production — and how to stop the fade?
• How data decisions shape every downstream fork in the journey?
• What “brains, behaviour, and nervous system” really mean in AI design?
• How to build hybrid platforms that stay compliant and fast?
• What does it take to shift from ad-hoc prompting to disciplined LLMOps?
• Why security, governance, and economics are the body’s immune system and heartbeat?

Srini breaks down the technical scaffold — how the ten cells of the honeycomb connect to deliver measurable ROI.

David probes rigorously from the business side — questioning trade-offs, accountability, and real-world friction.

Together they turn AI from a keynote fantasy into a hard-nosed operating reality.

AI dazzles in demos but stalls in the enterprise. In this deep dive with a new section on rapid-fire questions by Srini Annamaraju, the resident tech strategy expert on the No Effing AIdea podcast, David Royle , reveals why — and how a smarter Target Operating Model (T.O.M.) can finally bridge the gap between ambition and adoption.

This episode connects TOM blueprints to the people dimension: the rise of AI workflow architects, the tension between agents and humans, and the messy middle where governance meets innovation.

💡 What we cover:

• Why enterprises confuse AI projects for AI infrastructure
• How to design a T.O.M. that balances guardrails and greenlights
• The rise of new ‘hot’ AI roles — and which ones will fade fast
• What happens when test kitchens meet board-level control
• Why scaling AI isn’t a tech problem — it’s an operating model one

David brings creative clarity and a rare mix of strategic and operational design chops. Srini brings decades of enterprise GTM and technology experience.

Together, they unpack what it really takes to make AI run the enterprise — not just impress it.

Hosts Srini Annamaraju and David Royle are back for episode two – and yes, the feedback is in. Some said we were a bit too serious last time. We’ll try not to become a Sunday love songs show, but we’re working on upping the “gags per minute.”

This week’s conversation covers:

• Listener reach & feedback: Almost 100 plays already, with listeners tuning in from the UK, US, India, and even Slovenia. The appetite is real for discussions that go beyond hype and get into the messy middle of enterprise AI.
• Event notes from Big Data London: A buzzing show, but still very tech-heavy. We debate whether AI conversations are stuck in the IT lane, and why that’s a problem when the real impact is business-wide.
• NBER study on ChatGPT usage: 700M weekly users. Surprisingly, 70% of usage is personal rather than work. Heavy skew toward under-26s. We unpack what that means for adoption inside enterprises.
• US tech investment in the UK: Nvidia and OpenAI committing eye-watering sums (hundreds of billions over time). A rare bit of good economic news for the UK, with national implications for jobs, productivity, and independence from US/China dominance.
• Enterprise field news:
  • Citi experimenting with agentic AI for wealth advisors, using Claude and Gemini inside secure workspaces.
  • FT analysis showing CEOs hype AI on earnings calls, but get risk-heavy and muted in regulatory filings.
  • JLR cyberattack fallout: £3.5B revenue hit, no cyber insurance in place. Knock-on effects on suppliers and supply chain.
• The “three rooms” where AI decisions get made:
  • C-suite (value, governance, risk)
  • Technical teams (architecture, data quality, safe design)
  • Operations (ongoing management, compliance, usage quality)
• Traps to avoid:
  • The Whac-a-Mole Trap – hallucinations never disappear, they just reduce.
  • The Origami Trap – clever prompts aren’t a moat; without guardrails, they fold fast.
  • The IT-Only Trap – AI left to technologists will fail; business P&L owners need to lead.
  • The Corporate DNA Trap – over-automating risks erasing what makes your org unique.
• Shadow AI is real: Even if companies ban AI tools, staff use them on personal devices. Risks around leakage and compliance multiply.

We close with a look ahead:

• How frontier model labs (OpenAI, Cohere, Mistral, etc.) are approaching enterprise go-to-market.
• Real use cases from our own client work – what’s working, what’s not.

Next steps for listeners:
Got topics you’d like us to cover? Message us on LinkedIn. The more specific, the better – we’ll dig in and bring field notes to the next episode.

Episode 1 — No Effing AIdea! - AI ethics, job disruption, GenAI “failures,” the AI bubble, India SMB adoption, coding risks, consultants falling behind, biotech breakthroughs, and AI in education — all collide in our first episode of No Effing AIdea!

Welcome to the first episode. We are David Royle and Srini Annamaraju.

We set the tone with a stark cold open: new Stanford data shows a 13% drop in jobs for 22–25-year-olds in AI-vulnerable roles since ChatGPT launched. Then, in our Reality Check, we unpack the last two weeks of enterprise AI news with a fresh lens:

• MIT’s “95% failure” GenAI claim — and why that’s too simple.
• Why the so-called AI bubble might actually be good for business.
• Reliance & Meta’s $100M JV bringing enterprise AI to India’s SMBs.
• AI coding tools: 30% faster, but 2x more vulnerabilities.
• Big consultants left behind by in-house AI adoption.
• Stanford’s autonomous AI lab slashing drug discovery timelines.
• Khan Academy’s Khanmigo AI tutor bringing hope to 180M learners worldwide.

Our Deep Dive asks: why does AI suddenly get its own moral panic when cloud, ERP, and digital never did? We explore what “ethics” really means for enterprises today:

• How ethics shows up on the P&L — as fines, lawsuits, and PR disasters.
• Where ethics must live in the AI stack to avoid “governance theatre.”
• The trade-offs leaders underestimate — speed vs. trust, open vs. proprietary.
• A pragmatic three-step ethical readiness checklist for 2025.

Finally, in The Paradox Box, we tackle three dilemmas from the field:

• Boards demanding ROI and revolution at the same time.
• Compliance vs. engineers in the race for velocity.
• Customers rebelling against brilliance.

Listen in for pragmatic tactics, not theatre — and a candid take on why AI ethics isn’t an afterthought. It’s the seatbelt that lets enterprises drive faster.