¶ Introduction to AI Breakdown
Hi everyone. I'm Andy, and this is the AI Breakdown.
Welcome to your weekly news edition where I'll cover what happened in AI last week, why it matters all in less than 10 minutes. 4 00:00:14,353.457140008 --> 00:00:18,433.457140008 First up, the banks have gone all in in a way that would've felt unthinkable. 5 00:00:18,433.457140008 --> 00:00:22,243.457140008 Even a year ago, we saw three financial powerhouses. 6 00:00:22,483.457140008 --> 00:00:28,543.457140008 Wells Fargo, standard Chartered and Commonwealth Bank of Australia make enormous bets on ai. 7 00:00:28,963.457140008 --> 00:00:36,283.457140008 Firstly, Wells Fargo teamed up with Google Cloud to roll out AgTech AI across all 215,000 employees. 8 00:00:36,643.457140008 --> 00:00:42,403.457140008 So that's across every branch call center and the C-Suite with Google's agent space platform. 9 00:00:42,403.457140008 --> 00:00:48,343.457140008 At the core, over 2000 staff are already using it, but the aim is full enterprise coverage. 10 00:00:48,913.457140008 --> 00:00:56,508.457140008 Google Cloud described it as a defining moment for AgTech deployment in financial services and Wells Fargo's, CEO Charlie Shaw. 11 00:00:57,223.457140008 --> 00:00:59,833.457140008 Echo that momentum on an earnings call last month. 12 00:01:00,193.457140008 --> 00:01:06,883.457140008 He admitted it's still early days, but pointed to pilots running everywhere from branch systems to operations and call centers. 13 00:01:07,333.457140008 --> 00:01:09,883.457140008 Really across anywhere where you've got anything manual. 14 00:01:10,693.457140008 --> 00:01:14,473.457140008 The message is clear, even if the payoff hasn't fully landed yet. 15 00:01:14,893.457140008 --> 00:01:16,598.457140008 The bank's direction of travel is obvious. 16 00:01:17,278.457140008 --> 00:01:22,768.457140008 Anything manual from cool handling to back office tasks is heading for AI treatment. 17 00:01:23,398.457140008 --> 00:01:32,548.457140008 Meanwhile, standard Chartered launched their own gen AI S-C-G-P-T in 41 markets for upwards of 70,000 employees. 18 00:01:33,208.457140008 --> 00:01:34,618.457140008 This isn't just a chatbot. 19 00:01:34,828.457140008 --> 00:01:43,108.45714001 It's being used for sales, fraud, risk, software engineering, and customer support with local versions tailored for regions around the world. 20 00:01:43,828.45714001 --> 00:01:54,718.45714001 Their group CDO framed it as innovation with responsibility and the CEO called the adoption of Gen ai, an operational imperative for all banks down in Australia. 21 00:01:54,988.45714001 --> 00:02:05,938.45714001 Commonwealth Bank is going deep with open AI for fraud detection and personalization, building off a machine learning stack that already makes over 55 million daily decisions. 22 00:02:06,718.45714001 --> 00:02:08,683.45714001 And just to underline how much all this matters. 23 00:02:09,418.45714001 --> 00:02:10,348.45714001 These are banks. 24 00:02:10,438.45714001 --> 00:02:16,253.45714001 They're famously risk averse, regulated out the heirs, and traditionally five years behind everyone else. 25 00:02:16,903.45714001 --> 00:02:22,993.45714001 But here they are rolling out ai, not just as a feature, but as core essential infrastructure. 26 00:02:24,352.38645162 --> 00:02:28,432.38645162 Hot on the bank's heels is a seismic move in enterprise AI infrastructure. 27 00:02:28,822.38645162 --> 00:02:35,212.38645162 Google's now letting organizations run their Gemini models entirely on premises, thanks to the Google distributed cloud. 28 00:02:35,512.38645162 --> 00:02:44,632.38645162 So for anyone whose hair stands on end at the thought of putting sensitive data in the public cloud, think governments, insurers, or utilities, this is a big deal. 29 00:02:45,352.38645162 --> 00:02:53,302.38645162 The first mode is totally air gapped, which in non-tech speak means your AI runs on servers with no outside connection at all. 30 00:02:53,992.38645162 --> 00:03:02,572.38645162 Pilot customers in Singapore and Japan are already kicking the tires, and Google's also struck a partnership with Oracle to expand this capability even further. 31 00:03:03,202.38645162 --> 00:03:14,932.38645162 Simultaneously VMware now owned by Broadcom, has rolled out its own private AI toolkit building AI runtimes governance and security controls right into VMware Cloud Foundation. 32 00:03:15,562.38645162 --> 00:03:22,522.38645162 So enterprises can spin a private gen AI govern it centrally and use the same dashboard they already trust. 33 00:03:23,2.38645162 --> 00:03:26,692.38645162 They even added enhancements for securing AI agent traffic. 34 00:03:27,22.38645162 --> 00:03:29,812.38645162 Think AI bots talking to your internal databases. 35 00:03:30,142.38645162 --> 00:03:34,642.38645162 And introduced an intelligent assist tool to help IT teams automate tasks. 36 00:03:34,972.38645162 --> 00:03:45,772.38645162 Why does any of this matter? For years, highly regulated industries have been torn between the choice of moving data to the cloud and hope for the best on compliance or stick with legacy tools. 37 00:03:46,222.38645162 --> 00:03:51,532.38645162 Now, Google, VMware, and Oracle are saying you keep your data and get the latest ai. 38 00:03:52,117.38645162 --> 00:03:56,77.38645162 This shifts the competitive advantage as Broadcom's Perma Padman. 39 00:03:56,77.38645162 --> 00:03:58,867.38645162 Bon put it an AI app is just another app. 40 00:03:59,257.38645162 --> 00:04:03,787.38645162 We give you everything you need to deliver ag agentic apps quickly, safely, and with governance. 41 00:04:04,831.50287331 --> 00:04:07,831.50287331 Now it wouldn't be a weakened AI without some platform drama. 42 00:04:08,206.50287331 --> 00:04:18,706.50287331 Microsoft went public with its first two major in-house models, MAI voice one, which can pump out a minute of natural speech in under a second on a single GPU. 43 00:04:18,796.50287331 --> 00:04:19,966.50287331 That's seriously impressive. 44 00:04:20,206.50287331 --> 00:04:25,66.50287331 And may I one preview a general LM built to drive their co-pilot product line. 45 00:04:25,726.50287331 --> 00:04:34,96.50287331 Microsoft President of ai, musta fa Suleman said we have to be able to have the in-house expertise to create the strongest models in the world. 46 00:04:34,501.50287331 --> 00:04:35,671.50287331 The writing's on the wall. 47 00:04:35,881.50287331 --> 00:04:38,71.50287331 Microsoft isn't content to just be open. 48 00:04:38,191.50287331 --> 00:04:44,491.50287331 AI's biggest investor, they want to hedge, move faster and tailor products their own way over. 49 00:04:44,491.50287331 --> 00:04:47,521.50287331 At Meta, it's a very different but equally ambitious story. 50 00:04:48,31.50287331 --> 00:05:02,701.50287331 Meta struck a six year $10 billion cloud deal with Google for compute capacity, licensed image generation tech from Midjourney and signaled it might use OpenAI and Google's models as just another option in its consumer AI apps. 51 00:05:02,941.50287331 --> 00:05:08,371.50287331 At least while it preps its next Gen Llama models, A METAS spokesperson summed it up. 52 00:05:08,761.50287331 --> 00:05:14,971.50287331 We are taking an all of the above approach, building, world leading models, partnering and open sourcing. 53 00:05:15,556.50287331 --> 00:05:18,916.50287331 What's fascinating here is the stark contrast in approaches. 54 00:05:19,366.50287331 --> 00:05:22,276.50287331 Microsoft is doubling down on vertical integration. 55 00:05:22,636.50287331 --> 00:05:26,806.50287331 Building their own models gives them speed, control and independence from OpenAI. 56 00:05:27,436.50287331 --> 00:05:35,656.50287331 Meanwhile, Meta's taking a pragmatic hybrid strategy partnering where it makes sense while continuing their open source push. 57 00:05:35,656.50287331 --> 00:05:39,136.50287331 With Lama, both strategies acknowledge the same reality. 58 00:05:39,646.50287331 --> 00:05:44,476.50287331 AI is becoming too strategically important to rely solely on a single approach or partner. 59 00:05:45,932.98256071 --> 00:05:48,572.98256071 Let's put some hard numbers to the AI transformation. 60 00:05:48,962.98256071 --> 00:06:01,82.98256071 IBM's just published a survey showing 78% of organizations have adopted AI for at least one business function, and 83% expect AI agents to improve process efficiency by 2026. 61 00:06:01,502.98256071 --> 00:06:09,812.98256071 Even more striking, AI enabled workflows are set to jump from just 3% a day to a whopping 25% by the end of next year. 62 00:06:10,187.98256071 --> 00:06:11,417.98256071 That's an eightfold leap. 63 00:06:11,627.98256071 --> 00:06:15,677.98256071 Why is this actually important? It backs up everything we've just discussed. 64 00:06:16,127.98256071 --> 00:06:20,417.98256071 AI and business is no longer about experimentation or flashy pilots. 65 00:06:20,807.98256071 --> 00:06:25,127.98256071 It's being welded into the very processes and budgets that keep companies humming. 66 00:06:25,787.98256071 --> 00:06:30,887.98256071 Improved decision making was picked as the top benefit by nearly seven in 10 execs. 67 00:06:31,307.98256071 --> 00:06:35,957.98256071 If you're still on the sidelines, the competitive pressure to catch up is only going to rise. 68 00:06:36,527.98256071 --> 00:06:39,22.98256071 Honestly, I'm seeing a real shift on the ground. 69 00:06:39,707.98256071 --> 00:06:49,127.98256071 The teams moving fastest on tech unicorns, but well run disciplined businesses, automate and routine decisions, and unlocking major productivity. 70 00:06:49,847.98256071 --> 00:06:56,327.98256071 Workflow bottlenecks aren't just being eased, they're being reimagined by AI agents that work 24 7. 71 00:06:56,777.98256071 --> 00:07:03,557.98256071 The challenge is quickly becoming less about whether to use AI and more about where and how deep to embed it. 72 00:07:04,872.18534089 --> 00:07:08,952.18534089 Finally, let's talk about the one factor that isn't powered by silicon humans. 73 00:07:09,642.18534089 --> 00:07:14,232.18534089 A new Wiley survey shows that AI change fatigue is hitting the workforce hard. 74 00:07:14,712.18534089 --> 00:07:22,782.18534089 Two thirds expect the pace of change to increase, and over a third already reports severe stress from the unending parade of new tools and processes. 75 00:07:23,502.18534089 --> 00:07:27,282.18534089 Interestingly, most staff are still curious and excited about ai. 76 00:07:27,582.18534089 --> 00:07:32,682.18534089 44% say they use it weekly, but the gap is in support, clarity, and training. 77 00:07:33,192.18534089 --> 00:07:37,722.18534089 Only 18% feel they've received comprehensive training and a big chunk. 78 00:07:38,67.18534089 --> 00:07:40,617.18534089 Aren't even sure what their company's AI policy is. 79 00:07:41,487.18534089 --> 00:07:48,177.18534089 Middle managers are especially squeezed feeling responsible for delivering results while often lacking backing from leadership. 80 00:07:48,807.18534089 --> 00:07:49,137.18534089 Dr. 81 00:07:49,137.18534089 --> 00:07:56,397.18534089 Tracy Carney summed it up when employees are perpetually in reactive mode, change, fatigue, hurts productivity. 82 00:07:56,997.18534089 --> 00:08:09,207.18534089 So as business leaders, what should you be thinking about here? You need to clearly communicate your AI roadmap, give your managers the support they need, and roll out changes at a pace your people can actually absorb. 83 00:08:09,957.18534089 --> 00:08:12,477.18534089 Your success won't just depend on the technology. 84 00:08:12,717.18534089 --> 00:08:20,997.18534089 It'll hinge on whether real humans can keep up, push too hard, too fast, and you'll burn out your champions just as quickly as your resistors. 85 00:08:21,687.18534089 --> 00:08:28,827.18534089 And if you're planning to introduce new AI tools this quarter, remember this, invest as much energy in communication planning. 86 00:08:29,247.18534089 --> 00:08:34,527.18534089 Preparing for resistance and supporting your managers as you do on the shiny new tech itself. 87 00:08:34,887.18534089 --> 00:08:40,647.18534089 Because a resilient, well-trained team is your only guarantee that AI investments will actually pay off. 88 00:08:41,660.25655528 --> 00:08:45,260.25655528 So to wrap it up this week showed us two sides of the AI coin. 89 00:08:45,680.25655528 --> 00:08:51,830.25655528 On one hand, banks and enterprises are moving past experiments and putting AI right at the heart of their operations. 90 00:08:52,220.25655528 --> 00:08:59,240.25655528 On the other, we're reminded that the tech only works if people can keep up, stay trained, and actually trust the tools. 91 00:08:59,930.25655528 --> 00:09:05,90.25655528 The big takeaway, AI isn't just becoming core infrastructure, it's becoming core culture. 92 00:09:06,275.25655528 --> 00:09:13,220.25655528 If you're leading a team or a business, the winners won't just be the ones who plug AI in fastest, but those who bring their people along for the ride. 93 00:09:13,658.01165733 --> 00:09:15,608.01165733 That's all for this week's AI roundup. 94 00:09:15,968.01165732 --> 00:09:19,778.01165733 If you found value in this breakdown, please leave a rating and hit subscribe. 95 00:09:19,778.01165733 --> 00:09:20,708.01165732 See you next week.