Latest / Connecting the Dots / AWS Outage Hits Smart Beds, Copilot Gets Chatty, and IBM Brings Quantum to Wall Street

Transcript

AWS Outage Hits Smart Beds, Copilot Gets Chatty, and IBM Brings Quantum to Wall Street · Connecting the Dots

• 0:00
  Today's tech news moves fast, but understanding
• 0:03
  it shouldn't be hard. That's why Snarfl Solutions
• 0:06
  Group brings you Connecting the Dots, where we
• 0:09
  break down what's happening in tech and why it
• 0:11
  matters to you. Now, let's dive in with your
• 0:14
  hosts. Welcome back, everyone. It is Friday,
• 0:17
  October 24th, 2025. Good morning. Another great
• 0:21
  day here. Absolutely. And we've got some really
• 0:24
  interesting... tech shifts to unpack today is
• 0:27
  stuff that touches everything from, believe it
• 0:29
  or not, your mattress to, well, the future of
• 0:32
  finance. It's quite the range. And speaking of
• 0:35
  today, here in Sacramento, we're starting with
• 0:37
  some patchy fog, but that should burn off pretty
• 0:40
  quick. Mostly sunny skies expected high around
• 0:42
  72 degrees. Oh nice. Perfect fall weather lovely
• 0:45
  and Nationally just a quick heads up. There's
• 0:48
  a threat for some severe thunderstorms. Maybe
• 0:51
  flash flooding down the southern us Okay, and
• 0:53
  the Pacific Northwest is getting hit with one
• 0:55
  of those atmospheric rivers. So a bit more active
• 0:58
  elsewhere. Gotcha Well, hopefully things stay
• 1:00
  calm here. Now, how about the markets? We ended
• 1:02
  Thursday on a positive note, right? We did yeah
• 1:05
  Things look pretty upbeat. The Dow Jones finished
• 1:07
  up 282 .76 points. That's a solid 0 .60 % gain.
• 1:13
  Nice. And the S &P 500 also climbed at an 18
• 1:17
  .52 points, so up 0 .27%. Even Bitcoin got in
• 1:20
  on the action, up about $609, a 0 .55 % increase.
• 1:26
  So yeah, a pretty good end to the trading week,
• 1:28
  it seems. Good way to head into the weekend.
• 1:30
  Now, before we dive into today's stories, a quick
• 1:32
  shout out to our sponsor, Old Glory. Right. Old
• 1:35
  Glory. They're fantastic for music and sports
• 1:39
  fan merchandise. Really iconic stuff. And you
• 1:42
  know, if you're already thinking about holiday
• 1:44
  gifts, maybe for that techie friend who's hard
• 1:46
  to shop for. Yeah, definitely check them out.
• 1:48
  They have some genuinely fun and unique items.
• 1:50
  Head over to oldglory .com. Use the promo code
• 1:54
  SNARFUL. That's S -N -A -R -F -U -L at checkout.
• 1:58
  you'll get 15 % off your entire purchase. That's
• 2:00
  oldglory .com. Great deal. Okay. So let's get
• 2:04
  into our first story and this one. Well, it highlights
• 2:07
  some really kind of bizarre real world problems
• 2:10
  when our fancy tech depends completely on the
• 2:13
  cloud. Yeah. This involves that big AWS outage
• 2:15
  from earlier in the week. Amazon web services
• 2:17
  went down and it caused, uh, let's just say a
• 2:20
  nightmare for people with eight sleep pod systems.
• 2:22
  Right. These high -end smart mattress covers,
• 2:24
  they're not cheap, are they? Not at all. They
• 2:26
  start around $2 ,000, plus you have to pay a
• 2:28
  subscription like $17 a month or more just for
• 2:31
  the main features like temperature control. Wow.
• 2:34
  And the whole point is like perfect personalized
• 2:36
  sleep. That's the promise. But when AWS went
• 2:39
  down, those promises kind of evaporated. Because
• 2:42
  the beds need the cloud to work. Exactly. 100
• 2:45
  % reliant on cloud connectivity for commands.
• 2:48
  So when the servers went offline, the beds basically
• 2:51
  just froze. Or worse. Yeah, I saw some of the
• 2:55
  stories on Reddit and X. Pretty wild. Wild and
• 2:59
  frankly unacceptable for something marketed as
• 3:02
  health tech. We saw people saying their beds
• 3:05
  were stuck, fully inclined, you know, really
• 3:07
  uncomfortable positions. Okay, that's annoying.
• 3:09
  But it got worse, didn't it? Oh yeah. The temperature
• 3:11
  control failures were the really scary ones.
• 3:14
  Yeah. One person claimed their side of the bed
• 3:15
  just cranked itself up to 110 degrees Fahrenheit.
• 3:18
  What? 110? Yep. And they said it wouldn't turn
• 3:21
  down, just stuck there making their bed. Sweltering
• 3:25
  was the word they used. That's actually dangerous.
• 3:30
  Imagine paying all that money for something that
• 3:32
  tries to roast you because the server failed
• 3:33
  somewhere else. It really highlights the vulnerability,
• 3:36
  doesn't it? You buy this expensive hardware,
• 3:39
  but its basic function depends on an external
• 3:41
  service you have zero control over. And others
• 3:43
  had the opposite issue, like getting too cold.
• 3:46
  Right. Another user, someone who follows a strict
• 3:49
  no screens in the bedroom rule overnight, woke
• 3:52
  up freezing. Had to put on a sweater because
• 3:55
  the bed was stuck on a low temp. And without
• 3:57
  the app connecting to the cloud, they couldn't
• 3:59
  adjust it. Just basic comfort gone. So what happened?
• 4:03
  Did the company respond? They did. And this is
• 4:05
  maybe the silver lining. The CEO, Matteo Franceschetti,
• 4:08
  he apologized publicly and, you know, apparently
• 4:11
  responding to complaints they've had for years
• 4:13
  about this exact vulnerability. Finally listened.
• 4:16
  Seems like it. They quickly committed to and
• 4:18
  actually shipped an update called Outage Mode.
• 4:21
  Outage Mode. Okay, what does that do? It's the
• 4:23
  failsafe they should have had all along, really.
• 4:25
  It lets the app connect directly to the pod using
• 4:27
  Bluetooth locally. Ah! Bypassing the cloud entirely.
• 4:31
  Exactly. So even if AWS or their own servers
• 4:34
  go down, you can still do the basics. Yeah. Control
• 4:37
  the temperature, turn the thing on or off, flatten
• 4:40
  the base if it's inclined. It's kind of shocking
• 4:42
  that you wasn't there from day one, isn't it?
• 4:44
  For something that expensive and that... essential
• 4:46
  to your sleep. Totally agree. It's a huge lesson
• 4:49
  for the whole Internet of Things industry. If
• 4:52
  you're selling expensive hardware, especially
• 4:54
  for health or basic home functions, it needs
• 4:57
  local control, period. You're buying the device,
• 5:01
  not just renting access to a server. Well said.
• 5:04
  We need control over our beds and maybe even
• 5:06
  more so when AI starts remembering everything
• 5:08
  about us. Which brings us neatly to our second
• 5:11
  story. Right. Microsoft is giving its co -pilot
• 5:14
  AI a pretty major makeover. They're trying to
• 5:17
  make it more personal, more collaborative, almost
• 5:20
  like a teammate. Moving beyond just answering
• 5:22
  questions then. Way beyond. First big thing,
• 5:25
  co -pilot groups. Groups. Like multiple people
• 5:27
  using it together. Yeah. Up to 32 people. friends,
• 5:31
  study buddies, work colleagues, can all be in
• 5:33
  the same copilot session, collaborating with
• 5:35
  the AI. 32. That sounds like chaos. Huh? Yeah,
• 5:39
  maybe. Microsoft apparently thinks most groups
• 5:41
  will be smaller, like two or three people collaborating.
• 5:44
  OK, that makes more sense. And while it's rolling
• 5:46
  out for consumers first, you can immediately
• 5:48
  see the potential for work. Right. Imagine this
• 5:51
  built into Microsoft 365, the AI actively participating
• 5:55
  in your team meetings. Yeah, synthesizing notes,
• 5:58
  suggesting actions. I can see that. But you said
• 6:01
  more personal, too. Oh, yeah. They're adding
• 6:03
  an optional conversation style called real talk
• 6:05
  mode. Real talk? What does that mean? The AI
• 6:09
  gets sassy. Kind of. It's designed to adapt to
• 6:12
  your tone, but also importantly to add its own
• 6:14
  perspective. Its own perspective, so it'll argue
• 6:16
  with me. Well, maybe challenge assumptions is
• 6:19
  the term they use. It might offer different viewpoints,
• 6:21
  be witty, or even, yes, respectfully push back
• 6:25
  if it thinks information is incorrect. Huh, okay.
• 6:28
  That's a shift from assistant to sparring partner.
• 6:32
  Or maybe just a more engaged thought partner.
• 6:34
  That brings up trust, obviously. Yeah, if it's
• 6:37
  challenging me, I need to know why and I need
• 6:39
  to trust its reasoning. And how does memory fit
• 6:42
  in? That's the other huge piece. To be a real
• 6:44
  teammate, it needs memory. So Copilot is getting
• 6:48
  long term memory, meaning it will remember things
• 6:51
  about you across different conversations, appointments
• 6:53
  you mentioned, project details, maybe even like
• 6:56
  your anniversary if you told it. Whoa, OK, that
• 6:59
  sounds potentially useful, but also kind of creepy.
• 7:02
  How much does it remember? And who controls that?
• 7:05
  That's the crucial part. Microsoft seems very
• 7:08
  aware of the privacy concerns here. They're building
• 7:10
  in strong user controls. Like what? You can see
• 7:13
  a clear itemized list of everything Copilot remembers
• 7:17
  about you. No hidden data. Okay, transparency
• 7:20
  is good. And more importantly, you can delete
• 7:22
  things, delete individual facts, or even tell
• 7:25
  it using voice commands like, hey Copilot, forget
• 7:28
  everything you know about my partner. Really?
• 7:30
  Just tell it to forget stuff? Yep. Or forget
• 7:32
  my travel plans from last month. That level of
• 7:35
  granular control seems absolutely essential if
• 7:38
  we're going to be comfortable with AI getting
• 7:39
  this personal. Okay, that does make me feel a
• 7:42
  bit better. Control is key. And they're even
• 7:44
  giving it a face. Sort of. There's a new optional
• 7:47
  visual avatar called Miko. Miko! Like Clippy's
• 7:53
  grandchild. Please tell me it's not Clippy. No,
• 7:56
  they're describing it as a modern, human -centered
• 7:59
  character. It apparently shows expressions, reacts
• 8:02
  visually, and can act as a voice -enabled tutor
• 8:04
  in a new Learn Live mode. A tutor. Okay, so less
• 8:09
  annoying paperclip, more helpful animated guide.
• 8:12
  That seems to be the goal. It all points to this
• 8:14
  big shift, you know? AI isn't just a tool you
• 8:17
  call up anymore. It's becoming this persistent...
• 8:20
  personalized social partner. Fascinating. From
• 8:22
  cloud failed beds needing local control to AI
• 8:25
  needing memory control. Control seems to be the
• 8:27
  theme today. It really does. And that connects
• 8:29
  to our final story too, which is about bringing
• 8:31
  quantum computing closer to reality using surprisingly
• 8:35
  normal hardware. Quantum computing. OK, that
• 8:37
  always feels like science fiction, but IBM is
• 8:38
  making real progress. Big progress, yeah. It's
• 8:41
  less sci -fi, more practical steps now. The main
• 8:44
  challenge with quantum computers is that qubits,
• 8:47
  the basic information units, are super unstable
• 8:49
  and prone to errors. Right. Error correction
• 8:51
  is the big hurdle. Exactly. IBM's been working
• 8:54
  on complex algorithms to fix these errors, and
• 8:56
  the huge news is they successfully ran one of
• 8:59
  these key error -correcting algorithms in real
• 9:02
  time. Okay. And here's the kicker. They ran it
• 9:05
  on standard commercially available chips made
• 9:08
  by AMD, specifically FPGAs. FPGAs? Field programmable
• 9:13
  JIT arrays? I know those. They're not exotic
• 9:16
  quantum hardware, are they? Not at all. Think
• 9:18
  of them like flexible, reconfigurable chips.
• 9:20
  You can program them with software. They're way
• 9:23
  cheaper and easier to get than building custom
• 9:25
  quantum -specific silicon. So running quantum
• 9:27
  error correction on those chips, that's a big
• 9:29
  deal. It's a massive deal. IBM's quantum VP,
• 9:32
  Jay Gambetta, basically said, this proves the
• 9:34
  approach works on hardware that isn't ridiculously
• 9:37
  expensive. It lowers the cost barrier. Dramatically.
• 9:40
  Yeah. And it was fast, too, 10 times faster than
• 9:42
  they even needed it to be. This really accelerates
• 9:45
  their timeline for building a truly fault -tolerant
• 9:48
  quantum computer, which they're calling Starling,
• 9:51
  aiming for 2029. So using off -the -shelf -ish
• 9:54
  hardware makes that 2029 goal much more realistic.
• 9:58
  Precisely. It bypasses potentially years of waiting
• 10:02
  for specialized expensive hardware development.
• 10:05
  Okay, but is this still just lab stuff or is
• 10:07
  it doing anything useful today? That's the other
• 10:09
  exciting part. It's already showing real world
• 10:10
  value. They pointed to a recent project with
• 10:13
  the bank HSBC. Okay. They used IBM's current
• 10:17
  quantum processor combined with traditional computers
• 10:20
  what IDM calls their hybrid quantum -centric
• 10:23
  supercomputing approach. Hybrid, okay, so mixing
• 10:25
  quantum and classical. Right, and they applied
• 10:27
  it to a problem in algorithmic bond trading.
• 10:30
  Specifically, trying to better predict the likelihood
• 10:32
  that a particular trade would actually go through
• 10:35
  at the desired price. Which is super important
• 10:37
  in high -speed trading, right? Small edges matter
• 10:39
  a lot. Huge deal. And the results were impressive.
• 10:42
  Using the hybrid quantum classical method improve
• 10:45
  that prediction accuracy by up to 34 % Compared
• 10:48
  to just using the classical methods alone 34
• 10:51
  % Wow in finance. That's That's enormous. Yeah,
• 10:55
  HSBC actually called it a Groundbreaking world
• 10:59
  first in bond trading. It really validates IBM's
• 11:02
  hybrid strategy So you don't need the perfect,
• 11:05
  giant quantum computer of the future to get benefits
• 11:08
  now. You can start combining quantum strengths
• 11:10
  with today's computers. Exactly. It shows the
• 11:13
  path from experiment to actual commercial application
• 11:15
  is happening now. And IBM is positioning itself
• 11:18
  for what McKinsey thinks could be like a $198
• 11:21
  billion market by 2040. Incredible. From beds
• 11:24
  to AI teammates to quantum breakthroughs on standard
• 11:28
  chips. Quite a day for tech news. Definitely
• 11:30
  shows how fast things are moving. Now, before
• 11:32
  we wrap things up, one more quick mention for
• 11:34
  our other sponsor, Skoni. Yes, Skoni. If you're
• 11:37
  a golfer or no one, Skoni is the place for premium
• 11:40
  golf shoes. Seriously stylish, but also built
• 11:43
  for performance on the course. They look amazing.
• 11:45
  Check them out at skoni .com. That's S -K -O
• 11:49
  -N -I dot com. And just like with Old Glory,
• 11:52
  use the promo code SNARFL at checkout for 15
• 11:55
  % off your entire purchase. skoni .com promo
• 11:59
  code SNARFL. Great shoes. All right, so let's
• 12:01
  quickly recap the big threads from today. Okay.
• 12:04
  Story one, the eight sleep saga. A pretty stark
• 12:08
  warning about relying purely on the cloud for
• 12:10
  essential tech. Local control is absolutely necessary,
• 12:13
  right? Absolutely. Then story two, Microsoft
• 12:16
  copilot getting personal and social. AI is really
• 12:19
  shifting from just a tool to more of a partner,
• 12:21
  but thankfully with user control over that crucial
• 12:24
  memory aspect. That control piece is vital. And
• 12:26
  finally, IBM's quantum leap. running error correction
• 12:29
  on standard AMD chips and showing real -world
• 12:31
  results with HSBC. It just pulls quantum computing
• 12:34
  much closer to commercial reality, much faster
• 12:36
  than we thought. Three fascinating stories showing
• 12:38
  how tech is evolving and where the points of
• 12:41
  control really matter. Well put, lots to think
• 12:43
  about. We're here to help at Snrful, so reach
• 12:45
  out on our website if you have any more questions.
• 12:50
  That's a wrap on today's top stories. Remember,
• 12:54
  tech doesn't have to be complicated, and we're
• 12:56
  here to help connect the dots. If you've got
• 12:58
  questions or want to learn more, visit us at
• 13:01
  snarfle .com. Thanks for listening, and we'll
• 13:04
  catch you next time.