Latest / Today’s AI News / Anthropic Launches Claude Design, Three OpenAI Leaders Exit, an AI Song Hits Number One on iTunes

Transcript

Anthropic Launches Claude Design, Three OpenAI Leaders Exit, an AI Song Hits Number One on iTunes · Today’s AI News

• 0:00
  Welcome back to Today's AI News, brought to you by Nynex
• 0:02
  Productions. It is.
• 0:04
  It's great to have you with us again.
• 0:05
  Yeah. I'm glad to be here.
• 0:07
  So imagine hiring an architect who doesn't just sketch the
• 0:12
  blueprint for your dream house, but instantly builds the walls,
• 0:16
  paints the rooms, and wires the electricity while you watch.
• 0:20
  That's exactly what the software world is waking up to today.
• 0:23
  As the distance between an idea and a shipped product vanishes,
• 0:28
  we are watching the entire design stack collapse into a
• 0:31
  single chat window. We really are crossing a massive
• 0:34
  threshold here. I mean, the industry is rapidly
• 0:37
  abandoning the idea of, you know, jumping between 5
• 0:39
  different specialized apps in favor of unified instant
• 0:43
  execution environments. Exactly.
• 0:45
  And let's layout the road map for today's deep dive because
• 0:47
  there is a lot of ground to cover.
• 0:48
  We are unpacking Anthropic's aggressive move to essentially
• 0:52
  consume the software development stack, and that pressure is
• 0:55
  directly feeding into our second story, which is a massive
• 0:58
  leadership shakeup and radical restructuring over at Open AI.
• 1:03
  Then we're going to explore a flurry of rapid updates across
• 1:06
  the AI hardware and software landscape, including some, well,
• 1:10
  frankly some terrifying security implications.
• 1:12
  Yeah, that security stuff is a big deal.
• 1:14
  It really is. And finally, we have a truly
• 1:17
  fascinating look at how AI is being used down in New Zealand
• 1:21
  to engineer raw, acoustic folk music.
• 1:24
  Which is probably my favorite story today, honestly, just
• 1:26
  because it illustrates how these, you know, macro level
• 1:29
  multibillion dollar tech shifts are fundamentally altering the
• 1:33
  hyperlocal creative workflows of individual people.
• 1:36
  Absolutely, but we really have to start with the earthquake.
• 1:39
  In the design world, Anthropic just drop claw design, and this
• 1:42
  is powered by their new Opus 4.7 Vision model.
• 1:44
  Yeah, the pitch here is that it turns raw prompts, rough
• 1:47
  screenshots, or even your existing messy code bases into
• 1:51
  fully interactive prototypes, slide decks, and marketing
• 1:53
  materials. And the real breakthrough isn't
• 1:56
  just that it generates an image or a layout.
• 1:58
  I mean, we've had tools that do that for a while now, right?
• 2:00
  Right. The leap forward is how it
• 2:02
  manages contextual memory under the hood.
• 2:05
  Usually, if you want an AI to generate something that actually
• 2:09
  matches your company's aesthetic, you have to write
• 2:11
  this massive 50 page prompt detailing your specific hex
• 2:16
  codes, font hierarchies, corner radius.
• 2:18
  Every single time. Exactly every single time.
• 2:21
  But Claude Design bypasses that entirely during setup.
• 2:25
  It quietly digests your existing code bases, your previous mock
• 2:29
  ups, your GitHub repositories. It mathematically maps out your
• 2:33
  brand's DNA to build an auto apply system.
• 2:36
  So it essentially lives inside your brand's rulebook.
• 2:38
  You never have to remind it to use that specific corporate
• 2:41
  shade. Of blue precisely.
• 2:42
  It builds A persistent contextual memory, and the way
• 2:46
  you interact with it changes too.
• 2:48
  You aren't just crossing your fingers, hitting generate and
• 2:50
  hoping the prompt works perfectly on the first try.
• 2:52
  It is highly iterative. You refine the designs through a
• 2:55
  conversational chat or by dropping inline comments
• 2:58
  directly on the prototype. It even dynamically writes
• 3:01
  custom user interfaces for you on the fly.
• 3:03
  Wait, really? Custom interfaces.
• 3:05
  Yeah, say you tell it the layout feels too cramped.
• 3:09
  Instead of just guessing what you mean, the AI instantly
• 3:11
  generates a custom slider right there on your screen.
• 3:15
  You physically drag that slider to adjust the spacing until it
• 3:18
  looks right to your eye and the AI updates the underlying code
• 3:21
  in real time. That dynamic UI generation is
• 3:24
  wild, but to me, the ultimate killer feature here is the
• 3:27
  handoff. I mean, anyone who has ever
• 3:29
  built software knows the historic painful friction
• 3:33
  between the design team and the development team.
• 3:36
  Yeah, it's notorious. Right.
• 3:38
  A designer creates a beautiful pixel perfect mockup in a
• 3:41
  program like Figma, hands it over to the developer, and the
• 3:44
  developer immediately says, yeah, the physics of the web
• 3:47
  don't work like this. I can't build.
• 3:48
  This always. But with cloud design, the
• 3:51
  finish work goes straight into clawed code as a fully
• 3:54
  functioning build ready bundle. You can still export it to
• 3:57
  standard formats you know, like Canva, PPTX, PDF, or HTML if you
• 4:01
  really want to, but that clawed code integration is the death
• 4:06
  knell for the traditional translation step.
• 4:08
  It just does it. Exactly.
• 4:10
  The AI that designed the interface is directly handing
• 4:13
  the parameters over to the AI that writes the back end logic
• 4:17
  entirely within the same ecosystem.
• 4:19
  There is no longer a human developer trying to interpret a
• 4:23
  static image and turn it into React code.
• 4:26
  Which brings us to the corporate drama of it all.
• 4:30
  The timing of this launch was incredibly calculated.
• 4:32
  Unbelievably calculated. And Tropics Chief Product
• 4:35
  Officer Mike Krieger abruptly resigned from Figma's board of
• 4:39
  directors on April 14th. That was just three days before
• 4:42
  this Claude Design launch. It definitely removed any
• 4:44
  lingering doubt about anthropics ambitions.
• 4:47
  They are building a direct existential competitor to stand
• 4:50
  alone design giants like Figma. Wait, people are calling this
• 4:53
  just a design competitor? But that feels fundamentally
• 4:55
  wrong to me. Look at what Anthropic is
• 4:57
  actually piercing together. Here, yeah, it's bigger than
• 4:59
  that. They have cowork for team
• 5:01
  management, they have browser agents executing tasks, they
• 5:04
  have office suite integrations, and now Claude Design leading
• 5:08
  seamlessly into Claude code. This feels less like a
• 5:12
  standalone application and more like a total operating system.
• 5:14
  A black hole? Really.
• 5:15
  Exactly, it's a black hole pulling in every layer of the
• 5:18
  software stack. If Anthropic owns the ideation
• 5:22
  phase, the design phase, the coding phase, and the deployment
• 5:24
  phase, they aren't just selling you a better hammer.
• 5:28
  They've bought the entire hardware store and locked the
• 5:30
  doors from the inside. Is this signaling the death of
• 5:34
  standalone software apps? That is the multi billion dollar
• 5:38
  shift we are witnessing right now.
• 5:39
  The traditional Moat for a software company was having a
• 5:43
  highly specialized, hyper refined feature set that no one
• 5:46
  else could replicate. But when a foundational model
• 5:49
  like Opus 4.7 can dynamically generate the exact feature you
• 5:52
  need in real time, like that custom spacing slider we talked
• 5:55
  about, that traditional software mode evaporates.
• 5:58
  It just disappears. Completely.
• 6:00
  If an ecosystem takes you Florida State from a shower
• 6:02
  thought to a deployed application, the friction of
• 6:06
  paying for and switching between 5 different specialized tools
• 6:10
  becomes impossible to justify. They're competing with the
• 6:13
  concept of the traditional software suite itself.
• 6:16
  And that gravitational pull from Anthropic is placing A
• 6:20
  staggering amount of pressure on their competitors, which serves
• 6:23
  as the absolute perfect transition to the sheer chaos we
• 6:27
  are watching unfold over at Open AI.
• 6:30
  Chaos from the outside, perhaps, but potentially A harsh,
• 6:33
  calculated necessity from the inside.
• 6:35
  Well, let's lay out the facts for you listening In a single 24
• 6:38
  hour period, Open AI lost three massive senior executives.
• 6:42
  Yeah, a massive blow. We are talking about Kevin,
• 6:44
  while their former Chief Product Officer, Bill Peebles, the lead
• 6:47
  visionary on their video generation model, Sora and
• 6:50
  Srinivas Narayan, who is their Chief of Enterprise
• 6:52
  Applications. When you unpack each of those
• 6:54
  exits, they collectively tell the story of a massive corporate
• 6:57
  pivot. We'll take Kevin.
• 6:59
  While he was spearheading the Open AI for Science division,
• 7:03
  they were building dedicated standalone apps for researchers,
• 7:06
  like a tool called Prism. Right I.
• 7:08
  Remember that that entire division is now being
• 7:10
  decentralized. The science initiatives are
• 7:13
  being scattered and woven into their broader coding agent
• 7:16
  codex. It's no longer a stand alone
• 7:18
  product, it's just a feature within their enterprise suite.
• 7:21
  And then you have Bill Peebles. He was the face of Sora, the
• 7:25
  generative video model that blew everyone's minds a year ago.
• 7:28
  But Open AI completely killed the Sora app last month over
• 7:32
  cost. Yeah, that was shocking to a lot
• 7:35
  of people. Peebles released a statement
• 7:36
  calling his time there the honor and adventure of a lifetime, but
• 7:40
  the writing was on the wall. Generating hyper realistic video
• 7:43
  physics is astronomically expensive in terms of server
• 7:46
  compute, and open AI clearly decided it doesn't fit their
• 7:50
  core enterprise focus right now. Rounding out the three is
• 7:54
  Srinivas Narayan. He ran their enterprise apps for
• 7:56
  three years following a 13 year stint in the Facebook.
• 7:59
  His departure was framed around personal reasons, specifically
• 8:03
  moving to India to care for aging parents.
• 8:05
  But having all three of these foundational pillars drop on the
• 8:09
  exact same day sends a massive shockwave through the industry.
• 8:13
  Yeah, and Sam Altman tried to get ahead of the narrative with
• 8:15
  a blog post. He wrote that Open AI is now a
• 8:18
  major platform, not a scrappy startup, and that they need to
• 8:23
  operate in a more predictable way.
• 8:25
  The stated strategy is that they are killing what they view as
• 8:28
  side quests to catch up to Anthropic.
• 8:31
  Which is a very deliberate choice of words.
• 8:33
  But wait, Altman is calling this predictable.
• 8:36
  You don't just kill a massive flagship project like Sora,
• 8:39
  which was basically their primary marketing tool for six
• 8:41
  months and hemorrhaged 3 top executives in 24 hours because
• 8:45
  things are going predictably. To me this looks like a full
• 8:47
  blown panic response to Anthropic eating their lunch in
• 8:50
  the enterprise sector. I mean, is this a mature focus
• 8:53
  or is it panic? It absolutely looks erratic and
• 8:56
  the optics are terrible, but if you look at the raw economics it
• 8:59
  might be the brutal definition of corporate maturity.
• 9:02
  Really. How so?
• 9:04
  When you are a scrappy startup flush with venture capital, your
• 9:08
  goal is to place a hundred different bets.
• 9:10
  You build a video generator, you build a science app, you build
• 9:13
  voice models just to see what captures the public imagination.
• 9:17
  But when your goal shifts to becoming the foundational IT
• 9:20
  infrastructure for Fortune 500 companies, you have to look at
• 9:23
  profit and loss margins. Sora was a compute heavy side
• 9:27
  quest. It burned millions of dollars to
• 9:30
  generate cool videos, but it wasn't driving core B2B
• 9:34
  subscriptions the way Anthropics coding and design agents are.
• 9:37
  That makes sense. Abandoning the shiny, expensive
• 9:39
  toys to reallocate that server power to codecs, their coding
• 9:43
  agent, is a ruthless financial decision to secure the boring,
• 9:46
  highly lucrative enterprise stack.
• 9:48
  So while Open AI is aggressively narrowing its focus and killing
• 9:51
  off the side quests, the rest of the AI industry is realizing
• 9:55
  that executing these tasks requires a massive expansion of
• 9:58
  both software protocols and physical hardware.
• 10:01
  Definitely. Let's dig into quick hits and
• 10:03
  map out what else is happening across the landscape today.
• 10:05
  The overarching theme here is the shift from chat bots to
• 10:09
  agents. We are no longer just talking to
• 10:12
  the AI, we are handing it the mouse and keyboard.
• 10:15
  Exactly. On the tooling front, codecs
• 10:17
  just received a massive update. It now features full computer
• 10:21
  use capabilities and an embedded in app browser.
• 10:24
  And Entropics Cloud Opus 4.7 is also pushing advanced agentic
• 10:29
  coding. Right.
• 10:30
  And Perplexity has launched a product they are calling the
• 10:32
  Perplexity Personal Computer, which is an agent orchestrator
• 10:36
  designed to navigate your local files and local.
• 10:38
  Applications. It's crucial to understand the
• 10:40
  mechanical difference here. A chat bot gives you a list of
• 10:43
  instructions on how to reorganize your files or write a
• 10:46
  script. But an agent does it for you.
• 10:48
  Exactly. An agent orchestrator actually
• 10:50
  takes control of your operating system, opens the folders, moves
• 10:53
  the files and executes the script on your behalf.
• 10:56
  It is moving from conversation to autonomous action.
• 10:59
  But shifting from chat bots to autonomous agents comes with a
• 11:02
  massive hidden bottleneck, and that is compute power.
• 11:06
  It takes so much juice. Every time an agent takes
• 11:08
  control of your browser to scrape a website or test code,
• 11:12
  it is burning through server space, right?
• 11:14
  Google sees this bottleneck coming, which is why they are
• 11:17
  aggressively moving to bypass third party chip makers.
• 11:20
  They are reportedly partnering with Marvel to build a custom
• 11:23
  TPU, A tensor processing unit, and a memory processing unit
• 11:27
  specifically designed for AI inference.
• 11:31
  This is a classic vertical integration play to stop
• 11:34
  bleeding money and cut their reliance on Broadcom.
• 11:37
  In the AI world, you have training which is teaching the
• 11:40
  model, and inference, which is the model actually answering a
• 11:43
  user's prompt or executing a task.
• 11:45
  Inference is where the ongoing everyday cost lives and.
• 11:48
  It ends up fast. Insanely fast at Google scale.
• 11:51
  Every time they rely on a third party chip designer like
• 11:54
  Broadcom to handle that data, they are losing a massive slice
• 11:57
  of their profit margin. By designing custom silicon with
• 12:01
  Marvell, they can physically optimize the hardware
• 12:03
  architecture specifically for the quirks of their own Gemini
• 12:06
  models, drastically driving down the cost per query.
• 12:09
  That hardware optimization pairs perfectly with what is happening
• 12:13
  on the software side. To make these agents run
• 12:15
  smoother, Salesforce just launched a product called
• 12:18
  Headless 360. This is huge for developers.
• 12:21
  Yeah, they're exposing their entire customer relationship
• 12:24
  platform via MCP, which stands for Model Context Protocol.
• 12:28
  They are also opening up AP IS and command line interfaces so
• 12:32
  coding agents can directly act on customer data.
• 12:35
  To understand why this is a big deal, think about how an AI
• 12:38
  agent normally interacts with a website.
• 12:40
  If it has to use the graphical user interface, it has to look
• 12:43
  at the screen visually, locate the Add Contact button and
• 12:47
  simulate a mouse click. Which is super fragile.
• 12:50
  Incredibly brittle. If Salesforce updates their
• 12:52
  website design and moves that button 2 pixels to the left, the
• 12:56
  AI breaks. MCP acts like a universal
• 12:59
  translator plug socket instead of forcing the AI to navigate
• 13:02
  the visual website. Headless 360 gives the agent
• 13:05
  direct code level access to the raw nervous system of the
• 13:07
  database. It could just instantly and
• 13:09
  reliably read and write data. Exactly.
• 13:12
  Nos Research is doing the exact same thing right now by
• 13:15
  introducing Tool Gateway, which allows their Hermes agent to
• 13:18
  plug into multiple services without developers having to
• 13:21
  duct tape a dozen different APIs together.
• 13:24
  But giving autonomous agents direct access to the nervous the
• 13:27
  system of our software has a terrifying dark side.
• 13:31
  We had a security warning today. Versal just disclosed a security
• 13:34
  breach. Yeah, this was scary.
• 13:37
  It started with a hacked third party AI tool that users had
• 13:40
  connected to their Google accounts, and it ultimately
• 13:43
  impacted a limited subset of Versal's customers.
• 13:46
  Think about the gravity of this for a second.
• 13:48
  When you use a new AI agent and click Sign in with Google or
• 13:52
  granted Oath permissions to access your e-mail, your cloud
• 13:55
  storage, or your private code repositories, you are handing
• 13:58
  over a skeleton key. Wow.
• 14:00
  Yeah. If that seemingly harmless AI
• 14:02
  startup gets compromised by a hacker, the hacker doesn't just
• 14:05
  get your chat history, they inherit all the permissions you
• 14:08
  granted to that AI. They bypass your passwords
• 14:11
  entirely. We are rapidly creating massive
• 14:13
  invisible attack surfaces because we want the convenience
• 14:16
  of an AI organizing our digital lives.
• 14:19
  It is a massive blind spot that the industry is frantically
• 14:23
  trying to patch, and the capabilities of these models are
• 14:26
  only accelerating. Let's look at models and culture
• 14:28
  for a second. Anthropic CEO Dario Amade just
• 14:32
  gave an interview to the Financial Times where he stated
• 14:35
  that open source models, and specifically Chinese models,
• 14:38
  will reach Mythos capabilities in just six to 12 months.
• 14:42
  Mythos is an internal industry benchmark for absolute
• 14:46
  state-of-the-art reasoning and problem solving.
• 14:48
  Amodey is basically saying the absolute peak of today's
• 14:51
  proprietary technology will be free and open to anyone with an
• 14:54
  Internet connection by next year.
• 14:56
  That feels like a brutal expiration date on having any
• 14:59
  kind of technological advantage. And how can a company like
• 15:02
  Anthropic or Open AI possibly build a sustainable Moat if the
• 15:06
  open source community is constantly breathing down their
• 15:08
  necks, just six months behind their best models?
• 15:11
  It is the defining dilemma of the current tech era.
• 15:14
  If the raw intelligence, the model itself, is inevitably
• 15:17
  commoditized and given away for free by open source developers,
• 15:21
  then the intelligence cannot be your Moat.
• 15:23
  So what is? The Moat has to be the
• 15:25
  ecosystem. It connects right back to what
• 15:26
  we discussed with clawed design. A massive enterprise doesn't pay
• 15:30
  Anthropic just because the Opus 4.7 model is smart.
• 15:35
  They pay Empropic because that model is deeply securely
• 15:38
  integrated into their company's code base, their daily
• 15:41
  workflows, and their compliment structures.
• 15:43
  Oh, I see. The intelligence is becoming a
• 15:45
  utility exactly like electricity.
• 15:48
  You don't build a sustainable business by claiming you own the
• 15:50
  concept of electricity. You build it by creating the
• 15:53
  most indispensable, deeply integrated appliances that run
• 15:57
  on it. That makes total sense.
• 15:59
  And Speaking of what we can build with that electricity,
• 16:01
  there's another milestone today on the cultural front.
• 16:04
  An AI artist named Inga Rose just hit #1 on the iTunes global
• 16:08
  charts with a pop track called Celebrate Me, where the music
• 16:11
  was entirely generated by the AI platform Suno and the lyrics
• 16:15
  were by a human. Number one, globally, that's a
• 16:17
  huge it. Really is, but I want to
• 16:19
  contrast that massive global pop success with a hyperlocal,
• 16:24
  deeply technical use of AI in the music industry from our
• 16:28
  community segment. This.
• 16:29
  Is the New Zealand story. Yes, we're looking at a workflow
• 16:32
  from David P Down in New Zealand, who runs a small
• 16:35
  recording studio specializing in acoustic, folk and roots music.
• 16:39
  The juxtaposition here is just brilliant.
• 16:41
  You have AI churning out chart topping synthetic global pop
• 16:45
  music, while David is using AI to engineer raw human acoustic
• 16:51
  instruments. Right, so David used ChatGPT to
• 16:54
  code a custom plug in style app based around Cuba's plug
• 16:58
  insurance. For you listening who might not
• 17:00
  be an audio engineer. Cubis is a digital audio
• 17:03
  workstation, essentially a massive digital mixing board
• 17:05
  inside your computer. And writing DSP or digital
• 17:09
  signal processing code for audio plugins requires complex
• 17:12
  mathematics to physically alter sound waves.
• 17:14
  It is no small feat. No, it requires a deep
• 17:17
  understanding of both advanced programming and the physics of
• 17:19
  audio engineering. So.
• 17:21
  David runs his raw acoustic mix through this AI app he built.
• 17:24
  The app analyzes the entire song, mathematically identifies
• 17:27
  each individual instrument, the acoustic guitar, the stand up
• 17:30
  bass, the vocals, and provide surgical EQ and editing
• 17:33
  suggestions for every single track to bring the project to a
• 17:36
  mastering ready state. It even generates A graphical
• 17:39
  user interface with visual waveforms and data tables so he
• 17:43
  can easily read the analysis. What is fascinating here is the
• 17:46
  mechanics of how the AI acts as an analytical second set of
• 17:50
  years in a complex acoustic mix. Frequencies constantly clash.
• 17:55
  How does that work? Well, the low resonant notes of
• 17:58
  an acoustic guitar might be occupying the exact same Sonic
• 18:01
  space as the bass guitar, making the whole song sound muddy and
• 18:05
  undefined to the human ear. The AI maps out the math of
• 18:08
  those sound waves and suggests microscopic frequency cuts to
• 18:12
  carve out specific space for each instrument.
• 18:14
  And it goes even deeper. David can load a reference track
• 18:17
  into his custom app, say, a professionally mastered folk
• 18:20
  song from the 19 70s whose specific Sonic warmth he wants
• 18:24
  to emulate. Oh, that's incredibly smart.
• 18:26
  Yeah, the AI mathematically maps the topography of that vintage
• 18:29
  track and advises David on the exact EQ curves required to
• 18:33
  match that vibe on his own recording.
• 18:35
  But David emphasizes A crucial human element here.
• 18:38
  The AI just gives him the technical map.
• 18:40
  It allows him to manually adjust the dials on his mixed board,
• 18:44
  preserving his own creative vision and human touch while
• 18:47
  ensuring the underlying mathematics of the audio are
• 18:50
  flawless. It is the perfect symbiosis.
• 18:52
  The AI isn't a magic button that just does the mix for him and
• 18:56
  strips away the soul of the track.
• 18:57
  It is simply pointing out the mathematical pitfalls on the map
• 19:00
  and letting him drive the car. I just love the beautiful irony
• 19:03
  of this workflow. We are using ultra modern
• 19:06
  bleeding edge generative algorithms to mathematically
• 19:09
  perfect the raw organic friction of human fingers sliding across
• 19:13
  acoustic guitar strings. Really grounds the technology in
• 19:16
  reality. It really does.
• 19:17
  When we talk about clawed consuming the software stacker
• 19:20
  executives fling open AI. It feels very abstract.
• 19:24
  But David in New Zealand using AI to perfect a folk song, that
• 19:28
  is the actual utility of this revolution.
• 19:30
  It proves that intelligence is truly becoming ambient.
• 19:34
  It is quietly weaving itself into the technical fabric of
• 19:37
  every single digital interaction and creative process we have.
• 19:40
  We have covered massive ground today.
• 19:42
  We've gone from anthropic turning chat windows into
• 19:45
  instant software factories to open AI radically restructuring
• 19:49
  to survive the enterprise wars, all the way to bespoke AI audio
• 19:53
  engineering. It's been a busy day.
• 19:55
  But I want to leave you with a final provocative thought to
• 19:57
  Mull over. We talked about Claude Design,
• 20:01
  auto generating polished brand systems.
• 20:03
  By analyzing existing human made code bases, we talked about
• 20:07
  David's AI plug in perfectly mapping the frequency vibe of
• 20:10
  classic human made reference tracks.
• 20:13
  But what happens to human creativity when the reference
• 20:15
  tracks and the design inspirations themselves are all
• 20:18
  generated by AI? If future AI models are learning
• 20:21
  what is mathematically perfect by studying other AI generated
• 20:24
  perfection, are we entering an endless sterile echo Chamber of
• 20:28
  machine aesthetics? That is the big question.
• 20:31
  Where? Does the messy, unpredictable
• 20:33
  spark of true human originality fit in when the machine already
• 20:38
  knows the optimal mathematical formula for a hit song or a
• 20:42
  beautiful website? It is the existential question
• 20:44
  of our generation and we are going to have to navigate it
• 20:47
  very, very soon. Definitely something to think
• 20:49
  about. If you enjoyed today's deep
• 20:51
  dive, please make sure to subscribe so you stay fully
• 20:54
  updated on this incredibly fast moving space.
• 20:56
  We'd also love it if you could take a second to rate the show
• 20:59
  with five stars. It really helps us out.
• 21:01
  Thank you so much for joining us an we will see you back for
• 21:03
  tomorrow's episode.