November 25, 2025 • 10 mins
This week we’re looking at AI moving from hype to hands-on.
We start with OpenAI and DoorDash running “AI Jam” workshops for over 1,000 small businesses across US cities – helping restaurant owners, accountants and retailers build AI tools they can actually use the next day. Meanwhile, the tech giants are doing the opposite of slowing down: nearly $90 billion raised in fresh bonds to fuel data centres, GPUs and cloud infrastructure.
Nvidia smashed expectations again (with numbers even stronger than headlines suggest), Model ML just banked $75m to automate investment banking grunt work, and a mid-tier London accounting firm cut a two-week task down to two hours using Gemini 2.5.
At the same time, Amazon is telling engineers to ditch third-party coding assistants for its in-house AI tool “Kiro”, and the UK government wants regional AI Growth Zones to make AI adoption easier for business – though the details are fuzzier than the headline.
We pull it all together through one lens: AI is embedding fast at both ends of the market – from small cafés drafting menus in a workshop, to hyperscalers dropping billions to keep the GPU taps flowing. If you're building or scaling with AI, this episode gives you a heads-up on what’s coming next.
Mark as Played
Transcript
Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
(00:00):
Hi everyone.
(00:00):
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 00:00:13,837.764710762 --> 00:00:32,647.764710762 Let's start with something that, honestly, I really like the look of Open AI and DoorDash of teamed up for what they're calling the small business AI Jam, a series of hands-on workshops for more than 1000 small business owners across five US cities, San Francisco, New York, Houston, Detroit, and Miami. 5 00:00:33,232.764710762 --> 00:00:46,612.764710762 Instead of another glossy launch video, they're literally putting mentors in a room with people who run restaurants, retail shops, legal practices, accounting firms, and helping them build at least one AI tool they can actually use the next day. 6 00:00:47,32.764710762 --> 00:00:56,692.764710762 We're talking really practical stuff, drafting better marketing, copy sorting inventory, automating staff scheduling, writing those menu descriptions you've been putting off. 7 00:00:57,202.764710762 --> 00:01:00,52.764710762 DoorDash is Elizabeth Jineen summed it up nicely. 8 00:01:00,532.764710762 --> 00:01:02,392.764710762 Running a small business is brutal. 9 00:01:02,812.764710762 --> 00:01:05,872.764710762 AI can't make it easy, but it should make it easier. 10 00:01:06,352.764710762 --> 00:01:19,132.764710762 And Crystal Hane from OpenAI called Small Businesses the building blocks of a healthy neighborhood, and says, if you put powerful AI tools in the hands, you help them punch above their weight and get the fair piece of the economic pie. 11 00:01:19,732.764710762 --> 00:01:23,302.764710762 For me, this is AI moving from keynote slides to Main Street. 12 00:01:23,782.764710762 --> 00:01:26,602.764710762 If you are a SaaS founder, it's also a bit of a tell. 13 00:01:26,842.764710762 --> 00:01:31,102.764710762 OpenAI is seeding usage and expectations at the very edge of the market. 14 00:01:31,687.764710762 --> 00:01:41,287.76471076 The next wave of customers you sell to may already have AI baked into how they write emails, manage bookings, or talk to customers because they built those workflows. 15 00:01:41,287.76471076 --> 00:01:57,37.76471076 In a workshop like this, the big question is whether there's follow through a day of training is brilliant, but these businesses will need support, guardrails, and in some cases handholding to keep those tools useful and safe still is a signal of where AI is going. 16 00:01:58,377.90895127 --> 00:02:01,647.90895127 Now, zooming out from small businesses to the absolute giants. 17 00:02:02,217.90895127 --> 00:02:06,507.90895127 Over the past few weeks, the big US tech and cloud players have gone on a borrowing spree. 18 00:02:06,897.90895127 --> 00:02:12,837.90895127 Reuters tallies nearly $90 billion in new bonds, aimed largely at AI and cloud infrastructure. 19 00:02:13,347.90895127 --> 00:02:16,17.90895127 Alphabet raised about $25 billion. 20 00:02:16,347.90895127 --> 00:02:18,147.90895127 Met uh, $30 billion. 21 00:02:18,507.90895127 --> 00:02:25,827.90895127 Oracle $18 billion, Amazon $15 billion with others like Verizon and Salesforce also piling in. 22 00:02:26,547.90895127 --> 00:02:30,567.90895127 Traditionally, these firms have funded growth from their own cash or equity. 23 00:02:31,47.90895127 --> 00:02:34,677.90895127 The fact they're now leaning this hard on public debt is a pretty loud signal. 24 00:02:35,37.90895127 --> 00:02:41,217.90895127 Building AI ready data centers, buying GPUs, wiring up new regions, it's all very capital intensive. 25 00:02:41,727.90895127 --> 00:02:45,927.90895127 On the plus side, if you're building SAS or internal tools, this is good news. 26 00:02:46,317.90895127 --> 00:02:49,282.90895127 It means the cloud providers are not quietly cooling on the eye. 27 00:02:49,352.90895127 --> 00:02:50,242.90895127 They're doubling down. 28 00:02:51,87.90895127 --> 00:02:57,747.90895127 Expect more capacity, new AI specific services, and probably aggressive discounts to lock you into their stacks. 29 00:02:58,407.90895127 --> 00:03:01,767.90895127 For me, this doesn't scream that the AI bubble is bursting. 30 00:03:02,97.90895127 --> 00:03:05,697.90895127 It says we are all in on AI and we'll worry about the bill later. 31 00:03:06,357.90895127 --> 00:03:13,257.90895127 Great for innovation in the short term, but if you are building on top of these platforms, keep an eye on long-term cost exposure. 32 00:03:13,707.90895127 --> 00:03:17,882.90895127 Cheap GPU time today doesn't guarantee cheap GPU time in three years. 33 00:03:19,573.78221007 --> 00:03:23,563.78221007 Speaking of infrastructure, Nvidia just gave the market a fresh data point. 34 00:03:24,13.78221007 --> 00:03:31,963.78221007 The latest quarterly results came out this week, and they smashed expectations again, roughly 66% year on year sales growth. 35 00:03:32,323.78221007 --> 00:03:38,23.78221007 With about two thirds of their revenue now coming from data center chips, the stock jumped several percent on the news. 36 00:03:38,23.78221007 --> 00:03:44,833.78221007 Before settling back CEO, Jensen Wong used the earnings call to SWAT away all the AI bubble chatter. 37 00:03:45,223.78221007 --> 00:03:47,383.78221007 His line was that from their vantage point. 38 00:03:47,698.78221007 --> 00:03:52,678.78221007 What they're seeing in orders and deployments looks very different to a fad that's about to pop. 39 00:03:53,68.78221007 --> 00:03:59,878.78221007 Why should you care if you're not trading Nvidia stock? Because NVIDIA is effectively the heartbeat of enterprise AI build out. 40 00:04:00,463.78221007 --> 00:04:11,593.78221007 When their data center business is on fire, it means cloud providers and big enterprises are still pouring money into GPUs for training and running models for anyone building AI products. 41 00:04:11,593.78221007 --> 00:04:21,43.78221007 That's a quiet reassurance that the underlying platform is still expanding more clusters, more capacity, more incentive for clouds to differentiate with better tooling. 42 00:04:21,313.78221007 --> 00:04:23,233.78221007 I'd still treat it as cautious optimism. 43 00:04:23,683.78221007 --> 00:04:26,323.78221007 One strong quarter and a relieved stock market. 44 00:04:26,323.78221007 --> 00:04:37,63.78221007 Don't cancel out the broader questions about whether all this CapEx turns into sustainable revenue, but if you were worried that the AI party was about to be shut down by investors, think again. 45 00:04:38,483.0852908 --> 00:04:51,53.0852908 Moving on to financial services model ml, a startup based in San Francisco with offices in New York and London has just raised $75 million to automate a big chunk of what junior investment bankers do. 46 00:04:51,563.0852908 --> 00:04:58,313.0852908 Their tools use generative if AI to draft pitch decks, spin up financial models, and crank through due diligence reports. 47 00:04:58,703.0852908 --> 00:05:03,833.0852908 Basically, a lot of the classic analyst grunt work investors in the round include Ft. 48 00:05:03,833.0852908 --> 00:05:09,353.0852908 Partners, QED and Y Combinator, which is a serious FinTech vote of confidence. 49 00:05:09,683.0852908 --> 00:05:15,803.0852908 Remember, this company is only about a year old and had already raised $12 million earlier in 2025. 50 00:05:16,223.0852908 --> 00:05:19,133.0852908 If they pull it off, deal teams could move much faster. 51 00:05:20,183.0852908 --> 00:05:29,3.0852908 Imagine turning round an m and a pitch in days instead of weeks, because the first pass on the deck, the comps and even parts of the model are drafted by ai. 52 00:05:29,423.0852908 --> 00:05:32,33.0852908 But you can also see why junior bankers are feeling twitchy. 53 00:05:32,633.0852908 --> 00:05:37,823.0852908 The state admission is to replace much of the grunt work done by investment bankers for partners. 54 00:05:37,823.0852908 --> 00:05:39,113.0852908 That sounds like efficiency. 55 00:05:39,533.0852908 --> 00:05:46,313.0852908 If you are the grad who thought you'd spend three years in Excel, it sounds like your apprenticeship route is being rewritten in real time. 56 00:05:47,18.0852908 --> 00:05:53,918.0852908 This is a strong domain where AI could land repeatable documents, data heavy slides, templated analysis. 57 00:05:54,368.0852908 --> 00:06:07,628.0852908 The banks that get this right free up humans for client conversations and complex judgment calls, the ones that just use it to cut headcount without rethinking workflows are storing up risk because someone still has to own the quality of the output. 58 00:06:08,921.46661243 --> 00:06:13,271.46661243 Back to the UK now with a story that puts some real numbers on AI productivity. 59 00:06:14,141.46661243 --> 00:06:21,971.46661243 Reuters profiled London accountancy firm Moore Kingston Smith, which uses a platform based on Google's Gemini 2.51 60 00:06:21,971.46661243 --> 00:06:34,151.46661243 team cut a fraud check report from around two weeks down to two hours by letting clients upload entire data sets for automated analysis For an accounting practice, that's not a small tweak, that's a different operating model. 61 00:06:34,451.46661243 --> 00:06:39,786.46661243 Faster client reporting, less backlog, more time for actual advice instead of wrestling with documents. 62 00:06:40,691.46661243 --> 00:06:51,851.46661243 Zoom this out to the wider services economy, law firms, consultants, corporate finance teams, and you start to see why the UK government is betting on AI to help reboot productivity. 63 00:06:52,511.46661243 --> 00:06:58,541.46661243 If the average knowledge worker can claw back even a fraction of that time, it changes margins and capacity. 64 00:06:59,111.46661243 --> 00:07:09,851.46661243 Of course, there are the usual caveats over reliance on AI output without proper checking is a real risk, and staff understandably worry about what this does to jobs in the medium term. 65 00:07:10,466.46661243 --> 00:07:13,406.46661243 As a concrete present day example, it's powerful. 66 00:07:13,916.46661243 --> 00:07:15,536.46661243 AI isn't just a lab toy. 67 00:07:15,836.46661243 --> 00:07:19,706.46661243 It's quietly shortening month end from a slog to something manageable. 68 00:07:21,134.55116793 --> 00:07:24,164.55116793 Speaking of the uk, there was also a big political move this week. 69 00:07:24,509.55116793 --> 00:07:37,889.55116793 The UK government has set out plans for multi-billion pound AI investment and the creation of AI growth zones, regional hubs tied to big data center and skills projects with a heavy emphasis on AI skills and SME support. 70 00:07:38,309.55116793 --> 00:07:41,279.55116793 The idea is to turn these areas into proper ecosystems. 71 00:07:41,939.55116793 --> 00:07:51,209.55116793 Universities, startups, big tech partners like Microsoft and Google, all working together and to get thousands of people trained up with the eye skills. 72 00:07:51,569.55116793 --> 00:07:58,289.55116793 There's also specific support earmarked for small and medium enterprises to buy tools and get advice on adoption. 73 00:07:59,9.55116793 --> 00:08:03,599.55116793 If you're running a company in or near one of these zones, this is worth watching closely. 74 00:08:04,79.55116793 --> 00:08:05,279.55116793 It could mean grants for pilots. 75 00:08:05,999.55116793 --> 00:08:12,269.55116793 Subsidized cloud credits or just a better pipeline of AI literate talent coming out of local programs. 76 00:08:12,599.55116793 --> 00:08:15,869.55116793 The mood in the UK business community is optimistic. 77 00:08:16,199.55116793 --> 00:08:24,29.55116793 People like the direction, especially the focus on skills and mid-sized firms, but there's healthy skepticism around execution. 78 00:08:24,719.55116793 --> 00:08:27,179.55116793 We've heard big AI funding numbers before. 79 00:08:27,509.55116793 --> 00:08:33,509.55116793 The real test is whether money actually reaches frontline teams rather than disappearing into consultancy slide decks. 80 00:08:33,899.55116793 --> 00:08:44,69.55116793 Still as a signal, it's clear the UK wants to be seen as an AI friendly economy, and that can only help if you're trying to sell AI enabled products into that market. 81 00:08:45,410.53590465 --> 00:08:59,600.53590465 From governments to the big cloud platforms, again, this time Amazon Reuters reported on an internal memo telling Amazon software engineers to use the company's own AI coding assistant Kiro instead of third party tools. 82 00:09:00,395.53590465 --> 00:09:04,685.53590465 Kro launched back in July and does the usual code completion and generation tricks. 83 00:09:05,15.53590465 --> 00:09:11,225.53590465 But the key line in the memo was essentially, we're not planning to support any more external AI development tools. 84 00:09:11,675.53590465 --> 00:09:18,905.53590465 So if you're an Amazon engineer, happily using an external model for coding help, the writing's on the wall, you're expected to move to Kro. 85 00:09:19,355.53590465 --> 00:09:20,945.53590465 This is part productivity play. 86 00:09:21,215.53590465 --> 00:09:22,745.53590465 Part strategic control. 87 00:09:23,45.53590465 --> 00:09:26,40.53590465 By funneling all that internal usage through their own assistant. 88 00:09:26,750.53590465 --> 00:09:28,430.53590465 Amazon gets more data to improve. 89 00:09:28,430.53590465 --> 00:09:36,680.53590465 Kiro keeps sensitive code away from rival ecosystems and strengthens its own AI stack for AWS customers down the line. 90 00:09:37,370.53590465 --> 00:09:42,470.53590465 Developers unsurprisingly have mixed feelings on places like Hacker News and Reddit. 91 00:09:42,800.53590465 --> 00:09:51,860.53590465 Some people see it as a classic big company, strong arm move that could force teams onto a tool that today might not be as good as the competition. 92 00:09:52,340.53590465 --> 00:09:56,60.53590465 Others shrug and say, look, of course, Amazon wants Amazon engineers. 93 00:09:56,390.53590465 --> 00:09:57,590.53590465 Using Amazon tools. 94 00:09:57,920.53590465 --> 00:10:04,220.53590465 The bigger pattern is that large tech firms are starting to insist on proprietary AI inside their walls. 95 00:10:05,503.70589367 --> 00:10:07,483.70589367 That's all for this week's AI roundup. 96 00:10:07,813.70589367 --> 00:10:11,563.70589367 If you found value in this breakdown, please leave a rating and hit subscribe. 97 00:10:11,863.70589367 --> 00:10:12,823.70589367 See you next week.
Hi everyone.
(00:00):
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 00:00:13,837.764710762 --> 00:00:32,647.764710762 Let's start with something that, honestly, I really like the look of Open AI and DoorDash of teamed up for what they're calling the small business AI Jam, a series of hands-on workshops for more than 1000 small business owners across five US cities, San Francisco, New York, Houston, Detroit, and Miami. 5 00:00:33,232.764710762 --> 00:00:46,612.764710762 Instead of another glossy launch video, they're literally putting mentors in a room with people who run restaurants, retail shops, legal practices, accounting firms, and helping them build at least one AI tool they can actually use the next day. 6 00:00:47,32.764710762 --> 00:00:56,692.764710762 We're talking really practical stuff, drafting better marketing, copy sorting inventory, automating staff scheduling, writing those menu descriptions you've been putting off. 7 00:00:57,202.764710762 --> 00:01:00,52.764710762 DoorDash is Elizabeth Jineen summed it up nicely. 8 00:01:00,532.764710762 --> 00:01:02,392.764710762 Running a small business is brutal. 9 00:01:02,812.764710762 --> 00:01:05,872.764710762 AI can't make it easy, but it should make it easier. 10 00:01:06,352.764710762 --> 00:01:19,132.764710762 And Crystal Hane from OpenAI called Small Businesses the building blocks of a healthy neighborhood, and says, if you put powerful AI tools in the hands, you help them punch above their weight and get the fair piece of the economic pie. 11 00:01:19,732.764710762 --> 00:01:23,302.764710762 For me, this is AI moving from keynote slides to Main Street. 12 00:01:23,782.764710762 --> 00:01:26,602.764710762 If you are a SaaS founder, it's also a bit of a tell. 13 00:01:26,842.764710762 --> 00:01:31,102.764710762 OpenAI is seeding usage and expectations at the very edge of the market. 14 00:01:31,687.764710762 --> 00:01:41,287.76471076 The next wave of customers you sell to may already have AI baked into how they write emails, manage bookings, or talk to customers because they built those workflows. 15 00:01:41,287.76471076 --> 00:01:57,37.76471076 In a workshop like this, the big question is whether there's follow through a day of training is brilliant, but these businesses will need support, guardrails, and in some cases handholding to keep those tools useful and safe still is a signal of where AI is going. 16 00:01:58,377.90895127 --> 00:02:01,647.90895127 Now, zooming out from small businesses to the absolute giants. 17 00:02:02,217.90895127 --> 00:02:06,507.90895127 Over the past few weeks, the big US tech and cloud players have gone on a borrowing spree. 18 00:02:06,897.90895127 --> 00:02:12,837.90895127 Reuters tallies nearly $90 billion in new bonds, aimed largely at AI and cloud infrastructure. 19 00:02:13,347.90895127 --> 00:02:16,17.90895127 Alphabet raised about $25 billion. 20 00:02:16,347.90895127 --> 00:02:18,147.90895127 Met uh, $30 billion. 21 00:02:18,507.90895127 --> 00:02:25,827.90895127 Oracle $18 billion, Amazon $15 billion with others like Verizon and Salesforce also piling in. 22 00:02:26,547.90895127 --> 00:02:30,567.90895127 Traditionally, these firms have funded growth from their own cash or equity. 23 00:02:31,47.90895127 --> 00:02:34,677.90895127 The fact they're now leaning this hard on public debt is a pretty loud signal. 24 00:02:35,37.90895127 --> 00:02:41,217.90895127 Building AI ready data centers, buying GPUs, wiring up new regions, it's all very capital intensive. 25 00:02:41,727.90895127 --> 00:02:45,927.90895127 On the plus side, if you're building SAS or internal tools, this is good news. 26 00:02:46,317.90895127 --> 00:02:49,282.90895127 It means the cloud providers are not quietly cooling on the eye. 27 00:02:49,352.90895127 --> 00:02:50,242.90895127 They're doubling down. 28 00:02:51,87.90895127 --> 00:02:57,747.90895127 Expect more capacity, new AI specific services, and probably aggressive discounts to lock you into their stacks. 29 00:02:58,407.90895127 --> 00:03:01,767.90895127 For me, this doesn't scream that the AI bubble is bursting. 30 00:03:02,97.90895127 --> 00:03:05,697.90895127 It says we are all in on AI and we'll worry about the bill later. 31 00:03:06,357.90895127 --> 00:03:13,257.90895127 Great for innovation in the short term, but if you are building on top of these platforms, keep an eye on long-term cost exposure. 32 00:03:13,707.90895127 --> 00:03:17,882.90895127 Cheap GPU time today doesn't guarantee cheap GPU time in three years. 33 00:03:19,573.78221007 --> 00:03:23,563.78221007 Speaking of infrastructure, Nvidia just gave the market a fresh data point. 34 00:03:24,13.78221007 --> 00:03:31,963.78221007 The latest quarterly results came out this week, and they smashed expectations again, roughly 66% year on year sales growth. 35 00:03:32,323.78221007 --> 00:03:38,23.78221007 With about two thirds of their revenue now coming from data center chips, the stock jumped several percent on the news. 36 00:03:38,23.78221007 --> 00:03:44,833.78221007 Before settling back CEO, Jensen Wong used the earnings call to SWAT away all the AI bubble chatter. 37 00:03:45,223.78221007 --> 00:03:47,383.78221007 His line was that from their vantage point. 38 00:03:47,698.78221007 --> 00:03:52,678.78221007 What they're seeing in orders and deployments looks very different to a fad that's about to pop. 39 00:03:53,68.78221007 --> 00:03:59,878.78221007 Why should you care if you're not trading Nvidia stock? Because NVIDIA is effectively the heartbeat of enterprise AI build out. 40 00:04:00,463.78221007 --> 00:04:11,593.78221007 When their data center business is on fire, it means cloud providers and big enterprises are still pouring money into GPUs for training and running models for anyone building AI products. 41 00:04:11,593.78221007 --> 00:04:21,43.78221007 That's a quiet reassurance that the underlying platform is still expanding more clusters, more capacity, more incentive for clouds to differentiate with better tooling. 42 00:04:21,313.78221007 --> 00:04:23,233.78221007 I'd still treat it as cautious optimism. 43 00:04:23,683.78221007 --> 00:04:26,323.78221007 One strong quarter and a relieved stock market. 44 00:04:26,323.78221007 --> 00:04:37,63.78221007 Don't cancel out the broader questions about whether all this CapEx turns into sustainable revenue, but if you were worried that the AI party was about to be shut down by investors, think again. 45 00:04:38,483.0852908 --> 00:04:51,53.0852908 Moving on to financial services model ml, a startup based in San Francisco with offices in New York and London has just raised $75 million to automate a big chunk of what junior investment bankers do. 46 00:04:51,563.0852908 --> 00:04:58,313.0852908 Their tools use generative if AI to draft pitch decks, spin up financial models, and crank through due diligence reports. 47 00:04:58,703.0852908 --> 00:05:03,833.0852908 Basically, a lot of the classic analyst grunt work investors in the round include Ft. 48 00:05:03,833.0852908 --> 00:05:09,353.0852908 Partners, QED and Y Combinator, which is a serious FinTech vote of confidence. 49 00:05:09,683.0852908 --> 00:05:15,803.0852908 Remember, this company is only about a year old and had already raised $12 million earlier in 2025. 50 00:05:16,223.0852908 --> 00:05:19,133.0852908 If they pull it off, deal teams could move much faster. 51 00:05:20,183.0852908 --> 00:05:29,3.0852908 Imagine turning round an m and a pitch in days instead of weeks, because the first pass on the deck, the comps and even parts of the model are drafted by ai. 52 00:05:29,423.0852908 --> 00:05:32,33.0852908 But you can also see why junior bankers are feeling twitchy. 53 00:05:32,633.0852908 --> 00:05:37,823.0852908 The state admission is to replace much of the grunt work done by investment bankers for partners. 54 00:05:37,823.0852908 --> 00:05:39,113.0852908 That sounds like efficiency. 55 00:05:39,533.0852908 --> 00:05:46,313.0852908 If you are the grad who thought you'd spend three years in Excel, it sounds like your apprenticeship route is being rewritten in real time. 56 00:05:47,18.0852908 --> 00:05:53,918.0852908 This is a strong domain where AI could land repeatable documents, data heavy slides, templated analysis. 57 00:05:54,368.0852908 --> 00:06:07,628.0852908 The banks that get this right free up humans for client conversations and complex judgment calls, the ones that just use it to cut headcount without rethinking workflows are storing up risk because someone still has to own the quality of the output. 58 00:06:08,921.46661243 --> 00:06:13,271.46661243 Back to the UK now with a story that puts some real numbers on AI productivity. 59 00:06:14,141.46661243 --> 00:06:21,971.46661243 Reuters profiled London accountancy firm Moore Kingston Smith, which uses a platform based on Google's Gemini 2.51 60 00:06:21,971.46661243 --> 00:06:34,151.46661243 team cut a fraud check report from around two weeks down to two hours by letting clients upload entire data sets for automated analysis For an accounting practice, that's not a small tweak, that's a different operating model. 61 00:06:34,451.46661243 --> 00:06:39,786.46661243 Faster client reporting, less backlog, more time for actual advice instead of wrestling with documents. 62 00:06:40,691.46661243 --> 00:06:51,851.46661243 Zoom this out to the wider services economy, law firms, consultants, corporate finance teams, and you start to see why the UK government is betting on AI to help reboot productivity. 63 00:06:52,511.46661243 --> 00:06:58,541.46661243 If the average knowledge worker can claw back even a fraction of that time, it changes margins and capacity. 64 00:06:59,111.46661243 --> 00:07:09,851.46661243 Of course, there are the usual caveats over reliance on AI output without proper checking is a real risk, and staff understandably worry about what this does to jobs in the medium term. 65 00:07:10,466.46661243 --> 00:07:13,406.46661243 As a concrete present day example, it's powerful. 66 00:07:13,916.46661243 --> 00:07:15,536.46661243 AI isn't just a lab toy. 67 00:07:15,836.46661243 --> 00:07:19,706.46661243 It's quietly shortening month end from a slog to something manageable. 68 00:07:21,134.55116793 --> 00:07:24,164.55116793 Speaking of the uk, there was also a big political move this week. 69 00:07:24,509.55116793 --> 00:07:37,889.55116793 The UK government has set out plans for multi-billion pound AI investment and the creation of AI growth zones, regional hubs tied to big data center and skills projects with a heavy emphasis on AI skills and SME support. 70 00:07:38,309.55116793 --> 00:07:41,279.55116793 The idea is to turn these areas into proper ecosystems. 71 00:07:41,939.55116793 --> 00:07:51,209.55116793 Universities, startups, big tech partners like Microsoft and Google, all working together and to get thousands of people trained up with the eye skills. 72 00:07:51,569.55116793 --> 00:07:58,289.55116793 There's also specific support earmarked for small and medium enterprises to buy tools and get advice on adoption. 73 00:07:59,9.55116793 --> 00:08:03,599.55116793 If you're running a company in or near one of these zones, this is worth watching closely. 74 00:08:04,79.55116793 --> 00:08:05,279.55116793 It could mean grants for pilots. 75 00:08:05,999.55116793 --> 00:08:12,269.55116793 Subsidized cloud credits or just a better pipeline of AI literate talent coming out of local programs. 76 00:08:12,599.55116793 --> 00:08:15,869.55116793 The mood in the UK business community is optimistic. 77 00:08:16,199.55116793 --> 00:08:24,29.55116793 People like the direction, especially the focus on skills and mid-sized firms, but there's healthy skepticism around execution. 78 00:08:24,719.55116793 --> 00:08:27,179.55116793 We've heard big AI funding numbers before. 79 00:08:27,509.55116793 --> 00:08:33,509.55116793 The real test is whether money actually reaches frontline teams rather than disappearing into consultancy slide decks. 80 00:08:33,899.55116793 --> 00:08:44,69.55116793 Still as a signal, it's clear the UK wants to be seen as an AI friendly economy, and that can only help if you're trying to sell AI enabled products into that market. 81 00:08:45,410.53590465 --> 00:08:59,600.53590465 From governments to the big cloud platforms, again, this time Amazon Reuters reported on an internal memo telling Amazon software engineers to use the company's own AI coding assistant Kiro instead of third party tools. 82 00:09:00,395.53590465 --> 00:09:04,685.53590465 Kro launched back in July and does the usual code completion and generation tricks. 83 00:09:05,15.53590465 --> 00:09:11,225.53590465 But the key line in the memo was essentially, we're not planning to support any more external AI development tools. 84 00:09:11,675.53590465 --> 00:09:18,905.53590465 So if you're an Amazon engineer, happily using an external model for coding help, the writing's on the wall, you're expected to move to Kro. 85 00:09:19,355.53590465 --> 00:09:20,945.53590465 This is part productivity play. 86 00:09:21,215.53590465 --> 00:09:22,745.53590465 Part strategic control. 87 00:09:23,45.53590465 --> 00:09:26,40.53590465 By funneling all that internal usage through their own assistant. 88 00:09:26,750.53590465 --> 00:09:28,430.53590465 Amazon gets more data to improve. 89 00:09:28,430.53590465 --> 00:09:36,680.53590465 Kiro keeps sensitive code away from rival ecosystems and strengthens its own AI stack for AWS customers down the line. 90 00:09:37,370.53590465 --> 00:09:42,470.53590465 Developers unsurprisingly have mixed feelings on places like Hacker News and Reddit. 91 00:09:42,800.53590465 --> 00:09:51,860.53590465 Some people see it as a classic big company, strong arm move that could force teams onto a tool that today might not be as good as the competition. 92 00:09:52,340.53590465 --> 00:09:56,60.53590465 Others shrug and say, look, of course, Amazon wants Amazon engineers. 93 00:09:56,390.53590465 --> 00:09:57,590.53590465 Using Amazon tools. 94 00:09:57,920.53590465 --> 00:10:04,220.53590465 The bigger pattern is that large tech firms are starting to insist on proprietary AI inside their walls. 95 00:10:05,503.70589367 --> 00:10:07,483.70589367 That's all for this week's AI roundup. 96 00:10:07,813.70589367 --> 00:10:11,563.70589367 If you found value in this breakdown, please leave a rating and hit subscribe. 97 00:10:11,863.70589367 --> 00:10:12,823.70589367 See you next week.
Popular Podcasts
Stuff You Should Know
If you've ever wanted to know about champagne, satanism, the Stonewall Uprising, chaos theory, LSD, El Nino, true crime and Rosa Parks, then look no further. Josh and Chuck have you covered.
The Joe Rogan Experience
The official podcast of comedian Joe Rogan.
Two Guys, Five Rings: Matt, Bowen & The Olympics
Two Guys (Bowen Yang and Matt Rogers). Five Rings (you know, from the Olympics logo). One essential podcast for the 2026 Milan-Cortina Winter Olympics. Bowen Yang (SNL, Wicked) and Matt Rogers (Palm Royale, No Good Deed) of Las Culturistas are back for a second season of Two Guys, Five Rings, a collaboration with NBC Sports and iHeartRadio. In this 15-episode event, Bowen and Matt discuss the top storylines, obsess over Italian culture, and find out what really goes on in the Olympic Village.