<![CDATA[Ship AI]]>https://riverside.comRiverside.fm (https://riverside.com)Mon, 02 Nov 2026 14:13:55 GMTSun, 10 Aug 2025 20:26:37 GMT60Manav GuptaFrom 0 to Production. Practical tips, tricks, and best practices to make AI useful for production in the real world!episodicManav Guptamanavg@gmail.comno<![CDATA[ Sumeet Vaidya | AI Driven SDLC]]>Guest: Sumeet Vaidya, Co-founder & CEO, Crafting (previously Facebook, Uber, Discord)

Sumeet joins Manav to argue that the industry is optimizing the wrong half of the software lifecycle. Code generation is nearly free; validation is where the real engineering now lives.

What we cover:

The model debate is a distraction. Claude Code vs. Codex vs. Cursor matters tactically for an IC picking a tool today, but it's the wrong question for an engineering leader deciding where to invest. The right question is resilience — can your organization absorb the next wave without being locked in?

Resilience has two meanings. First, the boring one: if your primary provider goes down, does your engineering org stop? We learned this lesson with cloud providers and auth providers; AI infrastructure deserves the same treatment. Second, the harder one: if engineers increasingly don't touch code, how do you retain a deterministic way to evaluate and resolve failures in a probabilistic system?

Harnesses are mostly markdown. Sumeet's sharpest critique: the average "harness" is a pile of markdown files with no real contract, no true guardrails, and no answer for credentials, dependencies, or access. Loop engineering has the same gap — "check if it works" is doing enormous heavy lifting and getting glossed over.

Spec-driven development, honestly assessed. It serves the purpose architecture docs used to: forcing the orchestrator to refine the ask. It does not put hard guardrails on the agent. A beautiful spec that passes human scrutiny can still fall apart in execution. Useful, not a silver bullet.

The test taxonomy collapses. With the right infrastructure, the distinction between unit test, integration test, and CI/CD stops mattering — a test simply verifies something works, whether that spans one service or a hundred. Manav pushes on a pattern he sees repeatedly: engineers fooling themselves with high unit-test coverage while integration remains the weak point.

The compounding math. Even if loops cut regression probability in half, 10x more code shipped means roughly 5x more outages. Production-monitoring agents that detect and patch errors are another band-aid. Every step of that chain gets resolved upstream by real infrastructure investment.

Enterprise reality. Crafting deploys into customer clouds — no data leaves. Shared pre-production clusters, sanitized DB snapshots, credential injection via Vault, network interception for shared staging. Named customers include Brex, Fair, and Webflow; larger regulated logos are in flight. What enterprises actually ask about: security, visibility, and — increasingly — cost. Token maxing is blowing through budgets across the board.

Productivity, measured soberly. Not 100–200% gains. Double digits. Anyone counting PRs has lost the plot. Sumeet expects a winnowing of the middle layer of engineering, particularly outside Silicon Valley, alongside genuinely new demand — every dental office and clinic plausibly supports one engineer.

For new grads: CS isn't dead, it's changing. Someone who has relearned AI tooling four times in six months may be more adaptable than a senior engineer optimized for the old stack.

The closing question for your organization: When an agent's change passes every check and is still wrong in production — what's your blast radius, and who's accountable? If you can't answer cleanly, that's your roadmap.

]]>1fc527b2-307a-43c2-ba36-1b5d2d11f234Tue, 28 Jul 2026 14:00:00 GMT<p><b>Guest: Sumeet Vaidya, Co-founder &amp; CEO, Crafting</b> (previously Facebook, Uber, Discord)</p><p></p><p>Sumeet joins Manav to argue that the industry is optimizing the wrong half of the software lifecycle. Code generation is nearly free; validation is where the real engineering now lives.</p><p></p><p><b>What we cover:</b></p><p></p><p><b>The model debate is a distraction.</b> Claude Code vs. Codex vs. Cursor matters tactically for an IC picking a tool today, but it's the wrong question for an engineering leader deciding where to invest. The right question is resilience — can your organization absorb the next wave without being locked in?</p><p><b>Resilience has two meanings.</b> First, the boring one: if your primary provider goes down, does your engineering org stop? We learned this lesson with cloud providers and auth providers; AI infrastructure deserves the same treatment. Second, the harder one: if engineers increasingly don't touch code, how do you retain a deterministic way to evaluate and resolve failures in a probabilistic system?</p><p><b>Harnesses are mostly markdown.</b> Sumeet's sharpest critique: the average "harness" is a pile of markdown files with no real contract, no true guardrails, and no answer for credentials, dependencies, or access. Loop engineering has the same gap — "check if it works" is doing enormous heavy lifting and getting glossed over.</p><p><b>Spec-driven development, honestly assessed.</b> It serves the purpose architecture docs used to: forcing the orchestrator to refine the ask. It does not put hard guardrails on the agent. A beautiful spec that passes human scrutiny can still fall apart in execution. Useful, not a silver bullet.</p><p><b>The test taxonomy collapses.</b> With the right infrastructure, the distinction between unit test, integration test, and CI/CD stops mattering — a test simply verifies something works, whether that spans one service or a hundred. Manav pushes on a pattern he sees repeatedly: engineers fooling themselves with high unit-test coverage while integration remains the weak point.</p><p><b>The compounding math.</b> Even if loops cut regression probability in half, 10x more code shipped means roughly 5x more outages. Production-monitoring agents that detect and patch errors are another band-aid. Every step of that chain gets resolved upstream by real infrastructure investment.</p><p><b>Enterprise reality.</b> Crafting deploys into customer clouds — no data leaves. Shared pre-production clusters, sanitized DB snapshots, credential injection via Vault, network interception for shared staging. Named customers include Brex, Fair, and Webflow; larger regulated logos are in flight. What enterprises actually ask about: security, visibility, and — increasingly — cost. Token maxing is blowing through budgets across the board.</p><p><b>Productivity, measured soberly.</b> Not 100–200% gains. Double digits. Anyone counting PRs has lost the plot. Sumeet expects a winnowing of the middle layer of engineering, particularly outside Silicon Valley, alongside genuinely new demand — every dental office and clinic plausibly supports one engineer.</p><p><b>For new grads:</b> CS isn't dead, it's changing. Someone who has relearned AI tooling four times in six months may be more adaptable than a senior engineer optimized for the old stack.</p><p><b>The closing question for your organization:</b> When an agent's change passes every check and is still wrong in production — what's your blast radius, and who's accountable? If you can't answer cleanly, that's your roadmap.</p><p></p><p></p>no00:43:42 Sumeet Vaidya | AI Driven SDLCfull<![CDATA[John Capobianco | VibeOps, NetClaw & Network Automation]]>John is Head of AI & Developer Relations at Itential, a Google Developer Expert, former Cisco AI Technical Leader, former Senior Network Architect for the Parliament of Canada — and a fellow Canadian. He's the creator of NetClaw, an open-source AI agent that lets you talk to your network infrastructure in natural language, and the founder of VibeOps Forum, which grew from zero to 400+ members in weeks.

His thesis: after a decade of network automation evangelism, 70% of enterprise networks are still not meaningfully automated. But he no longer thinks they'll catch up the old way — because AI agents are changing the on-ramp entirely.

In this episode, Manav and John cover:

— Why VibeOps will eventually just become Ops (just like vibe coding became coding)

— How to build graduated trust with agents before giving them keys to your live network

— NetClaw: what it is, how it integrates with Cisco, Meraki, ACI, NetBox, Aruba and more

— How to generate a real-time 3D network topology using Blender via MCP

— What governed, enterprise-grade agentic ops actually looks like

— Spec-Driven Development: the next evolution after vibe coding

— John's bold prediction: by 2030, network engineers become HR managers for agents

If you lead infrastructure, run a network team, work in telco, or just want to understand what AI is about to do to IT operations — this episode is for you.

Find John: linkedin.com/in/johncapobianco | automateyournetwork.ca

NetClaw on GitHub: github.com/automateyournetwork/netclaw

Ship AI: https://manavgup.github.io/shipai/state-of-ai/

]]>f2f72cf0-1552-4c07-9f1e-6c6fd0c64086Tue, 21 Apr 2026 14:00:00 GMT<p>John is Head of AI &amp; Developer Relations at Itential, a Google Developer Expert, former Cisco AI Technical Leader, former Senior Network Architect for the Parliament of Canada — and a fellow Canadian. He's the creator of NetClaw, an open-source AI agent that lets you talk to your network infrastructure in natural language, and the founder of VibeOps Forum, which grew from zero to 400+ members in weeks.</p><p>His thesis: after a decade of network automation evangelism, 70% of enterprise networks are still not meaningfully automated. But he no longer thinks they'll catch up the old way — because AI agents are changing the on-ramp entirely.</p><p>In this episode, Manav and John cover:</p><p>— Why VibeOps will eventually just become Ops (just like vibe coding became coding)</p><p>— How to build graduated trust with agents before giving them keys to your live network</p><p>— NetClaw: what it is, how it integrates with Cisco, Meraki, ACI, NetBox, Aruba and more</p><p>— How to generate a real-time 3D network topology using Blender via MCP</p><p>— What governed, enterprise-grade agentic ops actually looks like</p><p>— Spec-Driven Development: the next evolution after vibe coding</p><p>— John's bold prediction: by 2030, network engineers become HR managers for agents</p><p>If you lead infrastructure, run a network team, work in telco, or just want to understand what AI is about to do to IT operations — this episode is for you.</p><p>Find John: <a rel="noopener noreferrer nofollow" href="http://linkedin.com/in/johncapobianco" target="_blank">linkedin.com/in/johncapobianco</a> | <a rel="noopener noreferrer nofollow" href="http://automateyournetwork.ca" target="_blank">automateyournetwork.ca</a></p><p>NetClaw on GitHub: <a rel="noopener noreferrer nofollow" href="http://github.com/automateyournetwork/netclaw" target="_blank">github.com/automateyournetwork/netclaw</a> </p><p>Ship AI: <a rel="noopener noreferrer nofollow" href="https://manavgup.github.io/shipai/state-of-ai/" target="_blank">https://manavgup.github.io/shipai/state-of-ai/</a></p>no00:58:33110John Capobianco | VibeOps, NetClaw & Network Automationfull<![CDATA[Alex Seymour & Kyle Sava | From Demo to Production: What Actually Breaks]]>Alex Seymour is a contributor to IBM's open-source agent infrastructure (BAI framework, Agent Stack, and the Relay project), exploring what it takes to build general-purpose agentic systems at enterprise scale.

Kyle Sava took a different path — he identified a real pain point as a tech seller, built a conversational AI roleplay tool on the side, and grew it into WatsonX Workshop: an internal AI-powered platform now being used by IBM's sales teams to practice pitches, prep for meetings, generate podcasts and videos, and learn products faster.

What we covered:

  • Why enterprise and consumer AI are more similar than you think — until governance enters the room
  • What "production-ready AI" actually means (Alex: it solves a problem. Kyle: it earns defensible trust at scale)
  • The first thing that breaks when you go from demo to production — hint: it's not what you think
  • The real state of MCP, A2A, and whether protocol standardization matters in enterprise
  • Why RAG doesn't scale vertically — and what to do about it
  • Advice for anyone deploying AI today: experiment relentlessly, and don't be afraid to let AI write the code
]]>803dc397-038d-49ee-b3dd-f65072f7ca62Tue, 14 Apr 2026 14:00:00 GMT<p><b>Alex Seymour</b> is a contributor to IBM's open-source agent infrastructure (BAI framework, Agent Stack, and the Relay project), exploring what it takes to build general-purpose agentic systems at enterprise scale.</p><p><b>Kyle Sava</b> took a different path — he identified a real pain point as a tech seller, built a conversational AI roleplay tool on the side, and grew it into <b>WatsonX Workshop</b>: an internal AI-powered platform now being used by IBM's sales teams to practice pitches, prep for meetings, generate podcasts and videos, and learn products faster.</p><p><b>What we covered:</b></p><ul><li>Why enterprise and consumer AI are more similar than you think — until governance enters the room</li><li>What "production-ready AI" actually means (Alex: it solves a problem. Kyle: it earns defensible trust at scale)</li><li>The first thing that breaks when you go from demo to production — hint: it's not what you think</li><li>The real state of MCP, A2A, and whether protocol standardization matters in enterprise</li><li>Why RAG doesn't scale vertically — and what to do about it</li><li>Advice for anyone deploying AI today: experiment relentlessly, and don't be afraid to let AI write the code</li></ul>no00:56:5819Alex Seymour & Kyle Sava | From Demo to Production: What Actually Breaksfull<![CDATA[Logan Kelly | Governing AI Agents]]>AI Governance Isn't an Afterthought — It's a Kill Switch

Logan Kelly, founder of Waxel, built AI agents for sales automation, shipped them to production, and immediately realized he had no control over cost, quality, or behavior.

That experience became the foundation for Waxell.ai — a control plane for governing enterprise agents.

Key takeaways from the conversation:

Observability alone is an autopsy. Tools like Datadog and Langsmith show you what went wrong after the fact. With agents, by the time you see it, data may already be exfiltrated, costs exploded, or cascading failures triggered across a chain.

The rug pull attack is real. MCP tool descriptions can be silently changed by a vendor — poisoning your agent's context and redirecting it to exfiltrate data or destroy records. Most enterprises have no detection for this.

Killing an agent is harder than it sounds. Stopping a runaway agent without losing state, audit trail, and replayability requires deliberate architecture — not just pulling the plug.

Multi-agent chains multiply risk exponentially. Every probabilistic handoff between agents compounds unpredictability. Governance has to be built in, not bolted on.

The missing piece in most AI strategies isn't more models — it's a unified governance layer that gives non-engineers visibility and control, regardless of which agents or frameworks are running underneath.

]]>a0d3f47f-9478-447a-a362-84ccd7121e1cTue, 07 Apr 2026 14:00:00 GMT<p><b>AI Governance Isn't an Afterthought — It's a Kill Switch</b></p><p></p><p>Logan Kelly, founder of Waxel, built AI agents for sales automation, shipped them to production, and immediately realized he had no control over cost, quality, or behavior. </p><p></p><p>That experience became the foundation for Waxell.ai — a control plane for governing enterprise agents.</p><p></p><p>Key takeaways from the conversation:</p><p></p><p><b>Observability alone is an autopsy.</b> Tools like Datadog and Langsmith show you what went wrong after the fact. With agents, by the time you see it, data may already be exfiltrated, costs exploded, or cascading failures triggered across a chain.</p><p></p><p><b>The rug pull attack is real.</b> MCP tool descriptions can be silently changed by a vendor — poisoning your agent's context and redirecting it to exfiltrate data or destroy records. Most enterprises have no detection for this.</p><p></p><p><b>Killing an agent is harder than it sounds.</b> Stopping a runaway agent without losing state, audit trail, and replayability requires deliberate architecture — not just pulling the plug.</p><p></p><p><b>Multi-agent chains multiply risk exponentially.</b> Every probabilistic handoff between agents compounds unpredictability. Governance has to be built in, not bolted on.</p><p></p><p><b>The missing piece in most AI strategies</b> isn't more models — it's a unified governance layer that gives non-engineers visibility and control, regardless of which agents or frameworks are running underneath.</p>no00:45:10Logan Kelly | Governing AI Agentsfull<![CDATA[Responsible AI by Design | Alex LaPlante, RBC]]>Alex LaPlante is VP of Cash Management Technology at RBC, former Interim Head of Borealis AI, co-author in Harvard Business Review, and a member of Canada's federal AI Strategy Task Force. In this conversation, she unpacks what separates organizations that actually ship AI from those stuck in demo mode — and why the answer is less about technology than culture, cross-functional collaboration, and asking the right questions before you build.

What we cover:

  • The "can we / should we" framework for every AI project
  • Why responsible AI has to be designed in from day one, not bolted on at the end
  • How RBC built an enterprise-grade MLOps platform to scale AI safely
  • Why "developer productivity" is the wrong frame — and what to measure instead
  • Agentic AI in regulated industries: the realistic deployment path
  • Canada's AI talent and commercialization gap — and what the Task Force is recommending
]]>47fc1c39-89c0-4cf6-b7a6-71c871032b78Tue, 31 Mar 2026 14:00:00 GMT<p>Alex LaPlante is VP of Cash Management Technology at RBC, former Interim Head of Borealis AI, co-author in Harvard Business Review, and a member of Canada's federal AI Strategy Task Force. In this conversation, she unpacks what separates organizations that actually ship AI from those stuck in demo mode — and why the answer is less about technology than culture, cross-functional collaboration, and asking the right questions before you build.</p><p></p><p><b>What we cover:</b></p><ul><li>The "can we / should we" framework for every AI project</li><li>Why responsible AI has to be designed in from day one, not bolted on at the end</li><li>How RBC built an enterprise-grade MLOps platform to scale AI safely</li><li>Why "developer productivity" is the wrong frame — and what to measure instead</li><li>Agentic AI in regulated industries: the realistic deployment path</li><li>Canada's AI talent and commercialization gap — and what the Task Force is recommending</li></ul>no00:46:3218Responsible AI by Design | Alex LaPlante, RBCfull<![CDATA[Episode 7 - AGI is coming]]>Seven Minutes to Midnight: AGI Is Coming

 

What is AGI? When is it arriving? And what does it mean for your career, your organization, and humanity?

In this episode of Ship AI, Manav Gupta delivers one of the most comprehensive, honest, and practical breakdowns of artificial general intelligence available today. No hype. No sci-fi. Just the data, the frameworks, and the hard questions.

Chapters

00:00 Introduction to AGI and Its Importance

13:56 The AGI Clock: Current Status and Predictions

17:49 The 10-Minute Problem in AGI Development

20:39 AI 2027 Timeline and Predictions

26:48 Agent Levels and Their Implications

30:34 Risk and Safety in AI Development

31:32 The Alignment Challenge in AI

36:34 The Five Walls: What Stands Between Current AI and AGI

41:07 Job Exposure: Technical Automatability vs Actual Displacement

44:03 Human Futures with AI

48:39 Future Scenarios for Humanity and AI

53:28 Navigating Career Paths in an AI World

Bottom line: Significant disruption is coming. The shape of it depends on decisions being made right now — by policymakers, by organizations, and by you.

If your core value is executing against known procedures, you are at risk. If your core value is judgment, trust, or physical presence, you have structural protection.

]]>29a8cfd9-ec96-4534-abec-18da45a1f90dTue, 17 Mar 2026 13:00:00 GMT<p><b>Seven Minutes to Midnight: AGI Is Coming</b></p><p> </p><p>What is AGI? When is it arriving? And what does it mean for your career, your organization, and humanity?</p><p></p><p>In this episode of Ship AI, Manav Gupta delivers one of the most comprehensive, honest, and practical breakdowns of artificial general intelligence available today. No hype. No sci-fi. Just the data, the frameworks, and the hard questions.</p><p></p><p><b>Chapters</b></p><p></p><p></p><p>00:00 Introduction to AGI and Its Importance</p><p>13:56 The AGI Clock: Current Status and Predictions</p><p>17:49 The 10-Minute Problem in AGI Development</p><p>20:39 AI 2027 Timeline and Predictions</p><p>26:48 Agent Levels and Their Implications</p><p>30:34 Risk and Safety in AI Development</p><p>31:32 The Alignment Challenge in AI</p><p>36:34 The Five Walls: What Stands Between Current AI and AGI</p><p>41:07 Job Exposure: Technical Automatability vs Actual Displacement</p><p>44:03 Human Futures with AI</p><p>48:39 Future Scenarios for Humanity and AI</p><p>53:28 Navigating Career Paths in an AI World</p><p></p><p><b>Bottom line:</b> Significant disruption is coming. The shape of it depends on decisions being made right now — by policymakers, by organizations, and by you.</p><p><i>If your core value is executing against known procedures, you are at risk. If your core value is judgment, trust, or physical presence, you have structural protection.</i></p>no01:00:1017Episode 7 - AGI is comingfull<![CDATA[Mihai Criveti, Context Forge]]>In this episode, Manav sits down with Mihai Criveti — IBM Distinguished Engineer and the creator of Context Forge, IBM's open-source agentic middleware — for a candid, technical conversation about what it actually takes to ship AI in the enterprise.

00:58 Navigating AI for Students

05:41 Context Forge and Scaling AI Agents

13:24 Understanding Context Forge

18:30 The Role of Middleware in AI

19:47 Evaluating AI Agents

24:10 Scaling AI in Enterprises

29:59 The Evolution of MCP Protocol

33:31 The Future of AI Agents and Workforce Dynamics

39:16 Lessons from Cloud Native to AI Agents

45:35 The State of AI and Future Predictions

]]>67258e75-0717-42a2-b10e-a00b2e57fdc6Tue, 10 Mar 2026 14:00:00 GMT<p>In this episode, Manav sits down with Mihai Criveti — IBM Distinguished Engineer and the creator of <b>Context Forge</b>, IBM's open-source agentic middleware — for a candid, technical conversation about what it actually takes to ship AI in the enterprise.</p><p></p><p></p><p>00:58 Navigating AI for Students</p><p>05:41 Context Forge and Scaling AI Agents</p><p>13:24 Understanding Context Forge</p><p>18:30 The Role of Middleware in AI</p><p>19:47 Evaluating AI Agents</p><p>24:10 Scaling AI in Enterprises</p><p>29:59 The Evolution of MCP Protocol</p><p>33:31 The Future of AI Agents and Workforce Dynamics</p><p>39:16 Lessons from Cloud Native to AI Agents</p><p>45:35 The State of AI and Future Predictions</p>no00:55:1523Mihai Criveti, Context Forgefull<![CDATA[Episode 6 - AI Agents.]]>The episode explores the rise of AI agents, their evolution from chatbots, and the challenges and opportunities in deploying and scaling AI agents. It delves into the characteristics of AI agents, the React pattern, advanced reasoning patterns, multi-agent orchestration, frameworks and protocols, governance, security implications, and the skills premium in the age of agency.

Takeaways

  • AI agents are evolving from chatbots to systems that think, plan, and act on behalf of users.
  • The React pattern, advanced reasoning patterns, multi-agent orchestration, and governance are critical components in the deployment and scaling of AI agents.

Chapters

  • 00:00 The Age of Agency
  • 15:33 Frameworks and Protocols
  • 30:17 Multi-Agent Orchestration
  • 47:47 Security Implications
]]>fcfba2d0-c1de-4754-81c9-dfab35cf18b4Fri, 06 Mar 2026 15:00:00 GMT<p>The episode explores the rise of AI agents, their evolution from chatbots, and the challenges and opportunities in deploying and scaling AI agents. It delves into the characteristics of AI agents, the React pattern, advanced reasoning patterns, multi-agent orchestration, frameworks and protocols, governance, security implications, and the skills premium in the age of agency.</p><p></p><p>Takeaways</p><ul><li>AI agents are evolving from chatbots to systems that think, plan, and act on behalf of users.</li><li>The React pattern, advanced reasoning patterns, multi-agent orchestration, and governance are critical components in the deployment and scaling of AI agents.</li></ul><p></p><p>Chapters</p><ul><li>00:00 The Age of Agency</li><li>15:33 Frameworks and Protocols</li><li>30:17 Multi-Agent Orchestration</li><li>47:47 Security Implications</li></ul>no00:53:2016Episode 6 - AI Agents.full<![CDATA[Ozge Yeloglu, VP AI & Analytics, CIBC]]>The conversation with Ozge Yeloglu covers her journey to becoming the VP of Advanced Analytics and AI at CIBC, her approach to deploying AI at scale, and the framework she built for success. It also delves into the concept of AI governance by design and the unique model of balancing AI governance and delivery, as well as leveraging the approach to change management for a large organization. 

Chapters:

  • 00:00 Introduction to Ozge Yeloglu and Her Journey
  • 08:53 AI Governance and Trustworthy AI Approach
  • 16:40 Building a Framework for AI Deployment
  • 29:49 People Change Management in AI
  • 38:20 Attracting and Retaining AI Talent
  • 43:17 The Future of AI: Hype vs. Reality
  • 49:38 Advice for Aspiring Engineers in AI
]]>18a66153-4c81-46ed-916b-1c42f809789bTue, 24 Feb 2026 14:00:00 GMT<p>The conversation with Ozge Yeloglu covers her journey to becoming the VP of Advanced Analytics and AI at CIBC, her approach to deploying AI at scale, and the framework she built for success. It also delves into the concept of AI governance by design and the unique model of balancing AI governance and delivery, as well as leveraging the approach to change management for a large organization. </p><p></p><p>Chapters:</p><ul><li>00:00 Introduction to Ozge Yeloglu and Her Journey</li><li>08:53 AI Governance and Trustworthy AI Approach</li><li>16:40 Building a Framework for AI Deployment</li><li>29:49 People Change Management in AI</li><li>38:20 Attracting and Retaining AI Talent</li><li>43:17 The Future of AI: Hype vs. Reality</li><li>49:38 Advice for Aspiring Engineers in AI</li></ul>no00:52:4524Ozge Yeloglu, VP AI & Analytics, CIBCfull<![CDATA[Episode 5 - AI Gets a Body]]>Episode Summary: When AI Gets a Body

This episode explores the paradigm shift from digital AI to physical AI—what happens when chatbots become robots. We trace the evolution from Karel Čapek's 1920 coining of "robot" to today's Cambrian explosion of humanoids, autonomous vehicles, drones, and brain-computer interfaces.


Chapters:
00:00 The Dawn of Physical AI

  1. 11:32 The $50 Trillion Opportunity
  2. 15:54 Basics of Robotics
  3. 28:07 Humanoids
  4. 37:56 Autonomous Vehicles
  5. 45:13 Defense AI
  6. 54:17 Space AI
  7. 57:22 Brain Computer Interface
  8. 01:07:42 The Human Dimension
  9. 01:11:15 Key Takeaways
]]>a8ece703-ac06-4928-b166-1b25005c7656Tue, 17 Feb 2026 14:00:00 GMT<p><b>Episode Summary: When AI Gets a Body</b></p><p>This episode explores the paradigm shift from digital AI to physical AI—what happens when chatbots become robots. We trace the evolution from Karel Čapek's 1920 coining of "robot" to today's Cambrian explosion of humanoids, autonomous vehicles, drones, and brain-computer interfaces.</p><p><br />Chapters:<br /><b>00:00 The Dawn of Physical AI</b></p><ol><li><b>11:32 The $50 Trillion Opportunity</b></li><li><b>15:54 Basics of Robotics</b></li><li><b>28:07 Humanoids</b></li><li><b>37:56 Autonomous Vehicles</b></li><li><b>45:13 Defense AI</b></li><li><b>54:17 Space AI</b></li><li><b>57:22 Brain Computer Interface</b></li><li><b>01:07:42 The Human Dimension</b></li><li><b>01:11:15 Key Takeaways</b></li></ol>no01:13:5015Episode 5 - AI Gets a Bodyfull<![CDATA[Lawrence Wan, Chief Architect, BMO]]>

Episode Overview

Lawrence Wan, Chief Architect and Innovation Officer at Bank of Montreal, shares insights on how one of Canada's largest financial institutions approaches technology transformation, AI adoption, and the future of agentic systems in a heavily regulated industry.

00:00 Introduction to Lawrence Van and BMO

05:01 The Role of Technology in Banking

11:19 Designing for Scale and Future Technologies

16:57 AI in Banking: Current Applications and Future Prospects

21:50 Generative AI: Productivity Gains and Challenges

28:22 The Future of AI Agents in Banking

33:34 The Evolving Role of Technologists in Banking

]]>c90fe12b-b1ae-4dd4-828d-c61e030d150bTue, 10 Feb 2026 14:00:00 GMT<p></p><p><b>Episode Overview</b></p><p>Lawrence Wan, Chief Architect and Innovation Officer at Bank of Montreal, shares insights on how one of Canada's largest financial institutions approaches technology transformation, AI adoption, and the future of agentic systems in a heavily regulated industry.</p><p></p><p></p><p>00:00 Introduction to Lawrence Van and BMO</p><p>05:01 The Role of Technology in Banking</p><p>11:19 Designing for Scale and Future Technologies</p><p>16:57 AI in Banking: Current Applications and Future Prospects</p><p>21:50 Generative AI: Productivity Gains and Challenges</p><p>28:22 The Future of AI Agents in Banking</p><p>33:34 The Evolving Role of Technologists in Banking</p>no00:39:2622Lawrence Wan, Chief Architect, BMOfull<![CDATA[AI In Banking - Vinh Tran, RBC]]>AI In Banking - Vinh Tran, VP Data & AI, RBC

In this episode, Vinh Tran—VP of Data and AI Platforms at RBC and an RBC Fellow—shares how one of the world's largest banks is approaching AI at enterprise scale.

Key Takeaways

"You have to control AI in order to scale it." Vinh explains that governance isn't about going slow—it's about building confidence. RBC invests heavily upfront in platforms, guardrails, and standardization so they can then open the doors and scale with assurance.

The Control Plane Approach: RBC has built a comprehensive AI control plane that includes:

  • A centralized LLM gateway controlling which models can be used (currently 6-8 rigorously validated models)
  • An MCP gateway for managing agent-to-tool connections with proper authentication
  • A controlled runtime environment for monitoring agent behavior at the transport level
  • An agent registry for inventory management and lifecycle control

00:00 Introduction to AI in Banking

02:01 The Role of AI in Enhancing Customer Experience

04:03 Governance and Compliance in AI Models

08:53 Control Plane: Ensuring Safe AI Deployment

14:53 Control Plane for Scaling AI

20:47 AI Guardrails

26:05 AI Agents and Standards

32:20 Predictions for the future

36:34 Advice for Aspiring Technologists

]]>e84ecff1-abc9-4ee8-bac3-60c73d4efb0bTue, 03 Feb 2026 13:00:00 GMT<p><b>AI In Banking - Vinh Tran, VP Data &amp; AI, RBC</b></p><p></p><p>In this episode, Vinh Tran—VP of Data and AI Platforms at RBC and an RBC Fellow—shares how one of the world's largest banks is approaching AI at enterprise scale.</p><p></p><p><b>Key Takeaways</b></p><p><b>"You have to control AI in order to scale it."</b> Vinh explains that governance isn't about going slow—it's about building confidence. RBC invests heavily upfront in platforms, guardrails, and standardization so they can then open the doors and scale with assurance.</p><p></p><p><b>The Control Plane Approach:</b> RBC has built a comprehensive AI control plane that includes:</p><ul><li>A centralized LLM gateway controlling which models can be used (currently 6-8 rigorously validated models)</li><li>An MCP gateway for managing agent-to-tool connections with proper authentication</li><li>A controlled runtime environment for monitoring agent behavior at the transport level</li><li>An agent registry for inventory management and lifecycle control</li></ul><p></p><p></p><p>00:00 Introduction to AI in Banking</p><p>02:01 The Role of AI in Enhancing Customer Experience</p><p>04:03 Governance and Compliance in AI Models</p><p>08:53 Control Plane: Ensuring Safe AI Deployment</p><p>14:53 Control Plane for Scaling AI</p><p>20:47 AI Guardrails</p><p>26:05 AI Agents and Standards</p><p>32:20 Predictions for the future</p><p>36:34 Advice for Aspiring Technologists</p>no00:40:3921AI In Banking - Vinh Tran, RBCfull<![CDATA[Episode 4 - AI At Work]]>The conversation explores the impact of AI adoption on the labor market, job tasks, disruption, and fluency. It delves into the Jevons Paradox, AI maturity gap, CEO mandates, labor market divergence, and productivity evidence, highlighting the reshaping of the labor market and the baseline expectation of AI proficiency at organizations.

Slides can be found at: https://manavgup.github.io/shipai/state-of-ai/ep04/1

Takeaways

  • AI adoption is reshaping the labor market
  • AI proficiency is now a baseline expectation at many organizations

Chapters

  • 00:00 The State of AI in the Labor Market
  • 00:00 Introduction to AI's Impact on the Job Market
  • 10:51 CEO Mandates for AI Proficiency
  • 15:58 The AI Maturity Gap in Companies
  • 21:14 The Reality of AI Adoption and Worker Sentiment
  • 28:40 Week-Long Action Plan for AI Adoption
  • 29:56 Wrapping Up
  • 31:45 Preview of Next Episode
]]>917e6f18-7fb0-40b7-a60d-a46f79f4bf27Mon, 02 Feb 2026 15:45:00 GMT<p>The conversation explores the impact of AI adoption on the labor market, job tasks, disruption, and fluency. It delves into the Jevons Paradox, AI maturity gap, CEO mandates, labor market divergence, and productivity evidence, highlighting the reshaping of the labor market and the baseline expectation of AI proficiency at organizations.</p><p></p><p>Slides can be found at: <a rel="noopener noreferrer nofollow" href="https://manavgup.github.io/shipai/state-of-ai/ep04/1" target="_blank">https://manavgup.github.io/shipai/state-of-ai/ep04/1</a></p><p></p><p>Takeaways</p><ul><li>AI adoption is reshaping the labor market</li><li>AI proficiency is now a baseline expectation at many organizations</li></ul><p></p><p>Chapters</p><ul><li>00:00 The State of AI in the Labor Market</li><li>00:00 Introduction to AI's Impact on the Job Market</li><li>10:51 CEO Mandates for AI Proficiency</li><li>15:58 The AI Maturity Gap in Companies</li><li>21:14 The Reality of AI Adoption and Worker Sentiment</li><li>28:40 Week-Long Action Plan for AI Adoption</li><li>29:56 Wrapping Up</li><li>31:45 Preview of Next Episode</li></ul>no00:32:3214Episode 4 - AI At Workfull<![CDATA[Episode 3 - The Red Silicon Curtain]]>Sanctions didn't kill Chinese AI—they mutated it into something more formidable: a leaner, inference-optimized, vertically-integrated competitor.

In this episode, we unpack how US export controls forced Chinese AI labs to innovate "up the stack," producing breakthroughs like DeepSeek's R1 model that matched frontier performance at a fraction of the cost. We explore China's $47.5 billion sovereign compute bet, the open-source war between Qwen, Llama, and Yi, and why China is deploying "good enough" humanoid robots at $16,000 while the West waits for AGI at $100,000+.

This is the story of how constraint creates innovation—and what it means for enterprise leaders navigating a bifurcating global AI landscape.

Chapters:

  • 00:00 The Mutation of Chinese AI
  • 05:15 DeepSeek's Disruption and Innovations
  • 15:53 The Open Source Revolution in AI
  • 29:04 China's National AI Strategy
  • 36:32 The Rise of Humanoid Robots
  • 46:29 Enterprise Implications and Strategic Questions

]]>3f6a813e-4187-4ba6-a596-01c8d4432dcaTue, 27 Jan 2026 12:00:00 GMT<p>Sanctions didn't kill Chinese AI—they mutated it into something more formidable: a leaner, inference-optimized, vertically-integrated competitor.</p><p>In this episode, we unpack how US export controls forced Chinese AI labs to innovate "up the stack," producing breakthroughs like DeepSeek's R1 model that matched frontier performance at a fraction of the cost. We explore China's $47.5 billion sovereign compute bet, the open-source war between Qwen, Llama, and Yi, and why China is deploying "good enough" humanoid robots at $16,000 while the West waits for AGI at $100,000+.</p><p>This is the story of how constraint creates innovation—and what it means for enterprise leaders navigating a bifurcating global AI landscape.</p><p></p><p></p><p>Chapters:<br /></p><ul><li>00:00 The Mutation of Chinese AI</li><li>05:15 DeepSeek's Disruption and Innovations</li><li>15:53 The Open Source Revolution in AI</li><li>29:04 China's National AI Strategy</li><li>36:32 The Rise of Humanoid Robots</li><li>46:29 Enterprise Implications and Strategic Questions</li></ul><p></p><p></p>no00:51:1213Episode 3 - The Red Silicon Curtainfull<![CDATA[Episode 2 - Follow the Money]]>In 2014, the largest tech companies spent $44 billion on capital investments. By 2024, that number passed $200 billion—almost all of it tied to AI. This isn't an innovation budget. This is the largest concentrated capital investment in corporate history.

In this episode, we follow the actual dollars: where they're going, why they're going there, and what has to be true for this multi-trillion dollar bet to pay off. We unpack the circular funding loops between OpenAI, Microsoft, NVIDIA, and Oracle. We examine why training costs 2.25x more than inference—and why that ratio has to flip. We explore the hidden taxes emerging around data licensing, regulatory compliance, and talent wars.

But here's what changes everything: nation states have committed over $500 billion to AI infrastructure since 2024. Saudi Arabia, UAE, France, South Korea—they're not playing by Silicon Valley rules. This is sovereign capital operating on political timelines, and it's providing a floor that reshapes the entire investment thesis.

Takeaways

  • AI infrastructure investments are driven by nation states and tech giants
  • The emergence of a circular funding ecosystem is reshaping the AI industry Small language models offer efficiency, speed, and accessibility, leading to productivity gains and cost savings for enterprises.
  • The AI industry faces challenges related to hidden costs, margin compression, regulatory compliance, and the impact of new entrants on the market.

Chapters

  • 00:00 Act 1 - The State of AI Investment
  • 08:48 AI Circular Economy
  • 17:24 $4T TAM
  • 19:42 Act 2 - The AI Machine
  • 25:39 Energy Constraints and Rise of Energy Industrial Complex
  • 28:23 Act 3 - The Hidden Costs
  • 37:11 Act 4- Business Model Crisis
  • 48:42 Act 5 - The Path Forward
  • 51:31 Nation States as New Players in AI
  • 54:22 Global AI Investment Landscape
  • 57:42 Emerging AI Business Models
  • 01:00:31 The Rise of Open Source AI
  • 01:03:20 Vertical AI Economics and New Models
  • 01:06:10 The Shift to Small Language Models

• • 01:09:03 Future Trends in AI Investment

]]>c5143ef3-f1a2-4c6d-bdc0-eed7bd84068cTue, 20 Jan 2026 12:30:00 GMT<p>In 2014, the largest tech companies spent $44 billion on capital investments. By 2024, that number passed $200 billion—almost all of it tied to AI. This isn't an innovation budget. This is the largest concentrated capital investment in corporate history.</p><p>In this episode, we follow the actual dollars: where they're going, why they're going there, and what has to be true for this multi-trillion dollar bet to pay off. We unpack the circular funding loops between OpenAI, Microsoft, NVIDIA, and Oracle. We examine why training costs 2.25x more than inference—and why that ratio has to flip. We explore the hidden taxes emerging around data licensing, regulatory compliance, and talent wars.</p><p>But here's what changes everything: nation states have committed over $500 billion to AI infrastructure since 2024. Saudi Arabia, UAE, France, South Korea—they're not playing by Silicon Valley rules. This is sovereign capital operating on political timelines, and it's providing a floor that reshapes the entire investment thesis.</p><p></p><p>Takeaways</p><ul><li>AI infrastructure investments are driven by nation states and tech giants</li><li>The emergence of a circular funding ecosystem is reshaping the AI industry Small language models offer efficiency, speed, and accessibility, leading to productivity gains and cost savings for enterprises.</li><li>The AI industry faces challenges related to hidden costs, margin compression, regulatory compliance, and the impact of new entrants on the market.</li></ul><p></p><p>Chapters</p><ul><li>00:00 Act 1 - The State of AI Investment</li><li>08:48 AI Circular Economy</li><li>17:24 $4T TAM</li><li>19:42 Act 2 - The AI Machine</li><li>25:39 Energy Constraints and Rise of Energy Industrial Complex</li><li>28:23 Act 3 - The Hidden Costs</li><li>37:11 Act 4- Business Model Crisis</li><li>48:42 Act 5 - The Path Forward</li><li>51:31 Nation States as New Players in AI</li><li>54:22 Global AI Investment Landscape</li><li>57:42 Emerging AI Business Models</li><li>01:00:31 The Rise of Open Source AI</li><li>01:03:20 Vertical AI Economics and New Models</li><li>01:06:10 The Shift to Small Language Models</li></ul><p>• • 01:09:03 Future Trends in AI Investment</p>no01:10:5912Episode 2 - Follow the Moneyfull<![CDATA[Episode 1 - The Speed of Now]]>The conversation delves into the exponential growth of AI models, the impact of compute abundance, global adoption of AI, and the comparison of AI performance with human capability. It explores the rapid evolution of AI technology and its implications on various aspects of society and industry.

All slides can be found at: https://manavgup.github.io/shipai/state-of-ai/ep01/1

Takeaways

  • Exponential Growth of AI Models
  • Impact of Compute Abundance
  • Global Adoption of AI
  • AI Performance vs. Human Capability AI performance is tightly coupled to the size of the training data
  • Frontier AI models are fundamentally data-driven
  • Compute investments into AI models determine their size and learning capability
  • AI progress is driven by compute, algorithms, and data
  • AI adoption is accelerating faster than any previous technology wave

Chapters

  • 00:00 The Speed of Now: Understanding AI's Rapid Evolution
  • 03:01 The Forces Behind AI's Acceleration
  • 05:47 The Impact of Moore's Law on AI
  • 08:31 The Role of Major Tech Companies in AI Growth
  • 11:27 Historical Context: 70 Years of AI Development
  • 14:38 NVIDIA's Rise and the End of Geographic Lag
  • 17:29 The Four Enablers of AI's Meteoric Rise
  • 20:20 Global Adoption and the New Reality of AI
  • 23:24 The Cambrian Explosion of AI Models
  • 26:17 AI Performance vs. Human Performance
  • 29:01 The Growth of Data Sets in AI Models
  • 29:47 The Exponential Growth of AI Data Sets
  • 32:58 The Role of Compute in AI Advancement
  • 36:42 Algorithmic Progress and Its Impact
  • 39:38 Benchmarking AI Performance Against Human Intelligence
  • 45:10 Enterprise Adoption of AI Technologies
  • 52:46 Key Learnings from AI Implementations
]]>9662ec7a-8ebf-49c6-84d0-ce489ebcd0dfTue, 13 Jan 2026 12:00:00 GMT<p>The conversation delves into the exponential growth of AI models, the impact of compute abundance, global adoption of AI, and the comparison of AI performance with human capability. It explores the rapid evolution of AI technology and its implications on various aspects of society and industry. </p><p></p><p>All slides can be found at: <a rel="noopener noreferrer nofollow" href="https://manavgup.github.io/shipai/state-of-ai/ep01/1" target="_blank">https://manavgup.github.io/shipai/state-of-ai/ep01/1</a></p><p></p><p>Takeaways</p><ul><li>Exponential Growth of AI Models</li><li>Impact of Compute Abundance</li><li>Global Adoption of AI</li><li>AI Performance vs. Human Capability AI performance is tightly coupled to the size of the training data</li><li>Frontier AI models are fundamentally data-driven</li><li>Compute investments into AI models determine their size and learning capability</li><li>AI progress is driven by compute, algorithms, and data</li><li>AI adoption is accelerating faster than any previous technology wave</li></ul><p></p><p>Chapters</p><ul><li>00:00 The Speed of Now: Understanding AI's Rapid Evolution</li><li>03:01 The Forces Behind AI's Acceleration</li><li>05:47 The Impact of Moore's Law on AI</li><li>08:31 The Role of Major Tech Companies in AI Growth</li><li>11:27 Historical Context: 70 Years of AI Development</li><li>14:38 NVIDIA's Rise and the End of Geographic Lag</li><li>17:29 The Four Enablers of AI's Meteoric Rise</li><li>20:20 Global Adoption and the New Reality of AI</li><li>23:24 The Cambrian Explosion of AI Models</li><li>26:17 AI Performance vs. Human Performance</li><li>29:01 The Growth of Data Sets in AI Models</li><li>29:47 The Exponential Growth of AI Data Sets</li><li>32:58 The Role of Compute in AI Advancement</li><li>36:42 Algorithmic Progress and Its Impact</li><li>39:38 Benchmarking AI Performance Against Human Intelligence</li><li>45:10 Enterprise Adoption of AI Technologies</li><li>52:46 Key Learnings from AI Implementations</li></ul>no00:59:4311Episode 1 - The Speed of Nowfull<![CDATA[Episode 1: GPT-5, Opus 4.1, GPT-OSS-2B, and more!]]>In this first episode, Manav Gupta and Mihai Criveti put the latest AI models through their paces in a head-to-head coding challenge. Watch as Claude Opus 4.1, GPT-5, and the open-source GPT OSS 20 billion compete to build interactive games and applications from simple prompts.

Highlights:

  • Live coding challenges including Snake, Minesweeper, and a Prince of Persia clone
  • Real-time comparison of how each model handles game development, from basic functionality to "kawaii" styling
  • Testing complex technical tasks like creating IBM mainframe architecture diagrams
  • Classic AI benchmark tests (counting letters, arithmetic problems) with surprising results
  • Mihai runs GPT OSS locally on his own GPU, showcasing impressive open-source capabilities

Key Takeaways:

  • Claude Opus 4.1 emerges as the overall winner with cleaner interfaces and superior artifact management
  • GPT-5 shows promise but struggles with canvas implementation
  • Open-source models are rapidly closing the gap with commercial offerings
  • Discussion on how these tools are reshaping the future of software development
]]>63b2d0e3-f519-45b4-8dfe-af4701322026Sun, 10 Aug 2025 20:28:36 GMT<p>In this first episode, Manav Gupta and Mihai Criveti put the latest AI models through their paces in a head-to-head coding challenge. Watch as Claude Opus 4.1, GPT-5, and the open-source GPT OSS 20 billion compete to build interactive games and applications from simple prompts.</p><p><b>Highlights:</b></p><ul><li>Live coding challenges including Snake, Minesweeper, and a Prince of Persia clone</li><li>Real-time comparison of how each model handles game development, from basic functionality to "kawaii" styling</li><li>Testing complex technical tasks like creating IBM mainframe architecture diagrams</li><li>Classic AI benchmark tests (counting letters, arithmetic problems) with surprising results</li><li>Mihai runs GPT OSS locally on his own GPU, showcasing impressive open-source capabilities</li></ul><p><b>Key Takeaways:</b></p><ul><li>Claude Opus 4.1 emerges as the overall winner with cleaner interfaces and superior artifact management</li><li>GPT-5 shows promise but struggles with canvas implementation</li><li>Open-source models are rapidly closing the gap with commercial offerings</li><li>Discussion on how these tools are reshaping the future of software development</li></ul>no00:40:0311Episode 1: GPT-5, Opus 4.1, GPT-OSS-2B, and more!full