Latest / Connecting the Dots / Connecting the Dots - 5-20-2025

Transcript

Connecting the Dots - 5-20-2025 · Connecting the Dots

• 0:00
  Today's tech news moves fast, but understanding
• 0:03
  it shouldn't be hard. That's why Snarful 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. Hello everyone and welcome back to Connecting
• 0:18
  the Dots. Hi there. It is Tuesday, May 20th,
• 0:22
  2025. We're your hosts, Alex and Morgan, coming
• 0:25
  to you from Snarful Solutions Group. Right here
• 0:28
  in sunny Sacramento, California. Yep. Definitely
• 0:31
  sunny today. Looks like a high of 88 degrees
• 0:33
  expected. Classic Sacramento spring. And speaking
• 0:37
  of highs, maybe a quick market check from yesterday's
• 0:40
  close. Monday, May 19th. Sure. Looks like the
• 0:42
  Dow Jones ended at 427 .92 .07. OK. And the S
• 0:47
  &P 500 closed up at 509 .63 .6. So pretty solid
• 0:52
  footing to start the week, market -wise. Indeed.
• 0:54
  But today, we're actually shifting focus away
• 0:57
  from the broad market. Yeah, we're doing a deep
• 0:59
  dive into the tech world. Exactly. We've got
• 1:01
  three pretty significant stories we want to unpack,
• 1:04
  things that aren't just headlines, but could
• 1:06
  actually impact how you navigate your digital
• 1:08
  life. Right. We're trying to connect the dots,
• 1:10
  basically, to see how these developments fit
• 1:12
  together, what they mean. OK, let's jump right
• 1:13
  into the first one. This involves a name many
• 1:15
  people will recognize. 23andMe. Ah, the genetic
• 1:20
  testing company. That's the one. So despite declaring
• 1:23
  bankruptcy, they've just been acquired, sold
• 1:26
  to Regeneron Pharmaceutical. Regeneron, the biotech
• 1:28
  firm. Okay. And the price tag. $256 million.
• 1:33
  Yeah. All happened in a bankruptcy auction. Wow.
• 1:36
  So bankruptcy, but still worth over a quarter
• 1:38
  billion. What's Regeneron actually getting for
• 1:41
  that? Well, they get nearly all of 23andMe's
• 1:44
  assets, the brand, the services, like the personal
• 1:47
  genome service. But the real value, I'd guess,
• 1:49
  is the data, right? That seems to be the consensus,
• 1:52
  the biobank. Genetic samples from over 15 million
• 1:55
  people, that's huge. 15 million, that's a massive
• 1:57
  data set for research. A genetic gold mine, really.
• 2:00
  So for people who use 23andMe or might still
• 2:03
  want to, what does this mean? Will the service
• 2:05
  continue? Yeah, Regeneron says the consumer -facing
• 2:08
  genome services, the DNA testing kits, all that
• 2:11
  will continue without interruption. Okay, continuity
• 2:13
  there, but the big question, the one everyone
• 2:15
  asks. What about the privacy of all that genetic
• 2:17
  data? It's incredibly sensitive stuff. Absolutely.
• 2:20
  That's the immediate red flag for a lot of people.
• 2:23
  And regulators are watching. The California Attorney
• 2:25
  General even urged folks to think about requesting
• 2:28
  their data be deleted. Right. There are probes
• 2:31
  into how the data has been handled, especially
• 2:32
  after. We'll get to the brooch. Yeah, so privacy
• 2:36
  is paramount here. What's Regeneron's position
• 2:39
  on this? Are they addressing the concerns? They
• 2:41
  are. They're emphasizing their commitment to
• 2:43
  data protection. They point to their own existing
• 2:46
  genetic center. which handles, I think, 2 .7
• 2:50
  million genomes already. OK, so they have experience
• 2:52
  with this scale of data. Right. Eris Baras, a
• 2:55
  senior VP there, specifically mentioned high
• 2:58
  standards for privacy and security for 23ME customers.
• 3:01
  And the president, George Yiannopoulos. He framed
• 3:04
  it as helping 23ME's mission while also boosting
• 3:07
  Regeneron's own large -scale genetics research.
• 3:10
  You know, finding new treatments, improving health.
• 3:12
  That's the pitch. It's quite a turn for 23ME,
• 3:14
  though. Weren't they valued at, like, $6 billion
• 3:16
  at one point? They were. Yeah, a huge valuation.
• 3:20
  But they really struggled to turn a profit. Their
• 3:22
  attempts at drug development didn't really take
• 3:24
  off. The subscription service didn't quite work
• 3:27
  out. And then there was the data breach. That
• 3:29
  had to hurt. Oh, massively. That was 2023. Affected
• 3:32
  nearly 7 million users. Exposed names, birth
• 3:36
  years, relationship labels, locations. Really
• 3:40
  damaging for trust. Yeah, that's tough to recover
• 3:42
  from when your whole business is built on people
• 3:45
  trusting you with their most personal data. Exactly.
• 3:48
  So this Regeneron deal, it feels like a big potential
• 3:51
  shift in the whole direct to consumer genetic
• 3:53
  space. Raises questions about, you know, the
• 3:56
  business models and who ultimately controls all
• 3:58
  this data. And it's not a done deal yet. No,
• 4:01
  it still needs approval in U .S. Bankruptcy Court.
• 4:03
  There's a hearing scheduled for June 17th. OK.
• 4:05
  And Regeneron plans to keep the 23andMe employees.
• 4:09
  Seems so. Yeah, but they are planning to shut
• 4:11
  down the lemonade health telehealth part of the
• 4:13
  business interesting. Okay a lot to watch there
• 4:15
  Let's pivot now second story. This one's for
• 4:18
  the developers mostly but has wider implications,
• 4:20
  right? Microsoft and the Windows subsystem for
• 4:24
  Linux WSL. Yeah, they've just open -sourced it
• 4:26
  put the code up on github which is pretty significant.
• 4:29
  For anyone unfamiliar with WSL, it came out back
• 4:32
  in 2016 with a Windows 10 update. It lets you
• 4:35
  run Linux command line tools, even graphical
• 4:37
  apps right inside Windows. It's been a huge deal
• 4:40
  for developers, right? It kind of broke down
• 4:42
  that wall between Windows and Linux. Absolutely.
• 4:45
  It made cross -platform work way easier. No more
• 4:48
  dual -booting half -holes or clunky VMs for many
• 4:50
  tasks. You get a real Linux environment within
• 4:53
  Windows. So why open source it now? What's the
• 4:56
  motivation? Well, according to Pierre Boulay,
• 4:58
  a senior software engineer at Microsoft, the
• 5:01
  community has already been super helpful with
• 5:02
  feedback and ideas. So this is about taking it
• 5:05
  a step further. Letting the community contribute
• 5:07
  code directly. Exactly. They're excited to see
• 5:09
  how WSL evolves with direct input, hoping it
• 5:12
  speeds things up, brings in new perspectives.
• 5:14
  OK, so what parts are actually open source? Is
• 5:17
  it everything? Almost. Key parts like the command
• 5:19
  line tools, WSL .exe, WSLG .exe, the background
• 5:24
  services, the Linux side daemons for networking
• 5:26
  and stuff. Those are all open now. That's not
• 5:28
  everything. No. A couple of core pieces remain
• 5:30
  closed source. The main kernel driver for the
• 5:33
  older WSL one, and some files handling the file
• 5:36
  system link between Windows and Linux. They say
• 5:38
  those are too tied into Windows itself. Got it.
• 5:41
  So the core Windows integration stays proprietary.
• 5:45
  but the Linux environment parts are open. Pretty
• 5:47
  much. Developers can now dig into that code,
• 5:50
  build it themselves, submit improvements, really
• 5:53
  understand how it works under the hood. What
• 5:55
  are the expected benefits then? Faster bug fixes.
• 5:57
  New features. All of the above, hopefully. Microsoft
• 6:00
  is looking for more community involvement in
• 6:02
  features, enhancements, fixes, more transparency,
• 6:05
  more collaboration. It should lead to a more
• 6:07
  robust toolset, better performance, potentially.
• 6:10
  Yeah, and better integration with Linux services.
• 6:12
  And apparently, this move actually closes out
• 6:14
  the very first issue ever filed on the WSL GitHub
• 6:18
  repo back in 2016. Wow, a nine -year -old request
• 6:21
  finally fulfilled through open source. That says
• 6:23
  something. It really does. It shows the long
• 6:26
  -term potential here for making WSL even better
• 6:29
  through community effort. It's good news for
• 6:31
  anyone doing development across Windows and Linux.
• 6:34
  Okay. Great development there. Let's move to
• 6:36
  our third and final story for today. This one
• 6:39
  looks ahead towards the future of AI. Right.
• 6:43
  Microsoft, again, they're adding support for
• 6:45
  something called the Model Context Protocol,
• 6:47
  or MCP. MCP. And that comes from Anthropic. Yep.
• 6:50
  The AI company. They created MCP late last year
• 6:54
  as an open standard. Think of it like a universal
• 6:57
  adapter for AI data. Like a USB -C port, but
• 7:00
  for AI models connecting to data sources. That's
• 7:03
  a great analogy. That's exactly the idea. Standardizing
• 7:06
  how AI models talk to external systems, cloud
• 7:09
  platforms like Azure or Google Cloud, CRMs, databases,
• 7:13
  whatever. So instead of building custom connections
• 7:14
  for every single data source? You'd use this
• 7:17
  single secure MCP interface. It's based on JSON
• 7:20
  RPC 2 .0. Jim Zemlin at the Linux Foundation
• 7:22
  even compared its potential role to what HTTP
• 7:25
  did for the web. OK, that's a big comparison.
• 7:27
  A foundational protocol for AI data exchange.
• 7:29
  Potentially, yeah. It solves a major headache.
• 7:32
  Getting AI systems to securely and reliably talk
• 7:35
  to all the different places data lives. And Microsoft
• 7:38
  is integrating this across their products. Pretty
• 7:41
  broadly. They mentioned GitHub, Copilot Studio,
• 7:45
  Dynamics 365, Azure and Azure AI services, Foundry
• 7:49
  agents, and even Windows 11 itself. Into the
• 7:52
  operating system. Interesting. What's the focus
• 7:54
  there? Security seems key. They're talking about
• 7:57
  enabling secure at -scale adoption of MCP. Which
• 8:01
  makes sense. Connecting AI to external data,
• 8:03
  that opens up risks. Huge risks. Microsoft specifically
• 8:07
  said they view input and training data for LLMs
• 8:10
  as inherently untrusted. Attacks on these connections,
• 8:13
  like on an MCP server, could, you know, jailbreak
• 8:16
  the AI or leak data. So how are they addressing
• 8:18
  that in Windows? They're planning security features
• 8:20
  like proxy mediated communication, tool level
• 8:23
  authorization, runtime checks. basically layers
• 8:25
  of security built into Windows 11 to manage these
• 8:28
  MCP connections safely. Okay. They also mentioned
• 8:30
  rebranding their AI platform inside Windows.
• 8:33
  Right. It's now going to be called Windows AI
• 8:35
  Foundry. Windows AI Foundry. So that's like the
• 8:40
  toolkit or environment for developers. building
• 8:42
  AI apps on Windows using these new capabilities.
• 8:45
  Seems like it. The foundation for AI within Windows.
• 8:48
  They also announced a new registry service for
• 8:51
  MCP servers within Windows, like a directory
• 8:53
  to find secure, trusted MCP endpoints. And there
• 8:57
  is another term, NLWeb? Yeah, NLWeb. They described
• 9:01
  it as an open project, comparing it to HTML,
• 9:04
  but for an agentic web. Agentic web, meaning
• 9:07
  a web designed for AI agents to interact with.
• 9:09
  That's the vision. Where every endpoint, every
• 9:12
  website maybe, could be an MCP server, making
• 9:14
  its content accessible and understandable to
• 9:17
  AI agents directly. It's a pretty forward -looking
• 9:19
  idea. It sounds ambitious. Are other major players
• 9:22
  involved in MCP? They are. Google and OpenAI
• 9:25
  have also joined the initiative, which... definitely
• 9:27
  gives it more weight, suggests it could become
• 9:29
  a real standard. What are the main benefits if
• 9:31
  MCP does take off like that? Well, unified integration,
• 9:34
  obviously, to a communication between AI and
• 9:36
  data, better scalability, consistency across
• 9:39
  systems, hopefully enhanced security because
• 9:42
  it's standardized, and less development time
• 9:44
  spent building custom connectors. Can you give
• 9:46
  some examples? Like, how would this be used?
• 9:48
  Sure. Think customer support chatbots that can
• 9:50
  pull real time info from any CRM system via MCP.
• 9:55
  Or... enterprise search tools that seamlessly
• 9:57
  search across documents, databases, cloud storage,
• 10:00
  developer tools that integrate AI assistance
• 10:02
  more deeply, even autonomous AI agents coordinating
• 10:06
  tasks across different services. Microsoft's
• 10:09
  bigger picture here is this agentic web, enabling
• 10:12
  AI agents from different companies to collaborate,
• 10:15
  almost like how websites link together now. Exactly.
• 10:17
  Creating that common ground using protocols like
• 10:20
  MCP. They also talked about improving AI memory
• 10:23
  using structured retrieval augmentation, basically,
• 10:26
  helping AI remember context better by connecting
• 10:28
  to these structured data sources. Building the
• 10:30
  plumbing for a more capable interconnected AI
• 10:32
  future. That definitely seems to be the goal.
• 10:35
  Making AI agents more powerful and useful by
• 10:38
  standardizing how they access and interact with
• 10:40
  the world's data. Wow. Okay, that's a lot to
• 10:43
  digest across those three stories. Let's try
• 10:45
  to quickly recap the main takeaways. That's good.
• 10:48
  First, we had 23Anime being acquired by Regeneron,
• 10:51
  raising really important questions about genetic
• 10:53
  data privacy even as the consumer service continues.
• 10:56
  Right. Big focus on data control there. Then
• 10:59
  Microsoft open -sourcing the Windows subsystem
• 11:01
  for Linux, WSL, a big move for developers, fostering
• 11:05
  more collaboration and transparency. And finally,
• 11:07
  Microsoft embracing Anthropix model context protocol,
• 11:11
  MCP, and launching the Windows AI Foundry. This
• 11:14
  is all about building a more standardized, secure,
• 11:17
  and interconnected future for AI agents. Yeah,
• 11:19
  laying the groundwork for how AI interacts with
• 11:22
  data systems. And each of these, you know, they
• 11:24
  touch on different aspects of our digital lives,
• 11:26
  data ownership, the tools developers use, and
• 11:29
  the future shape of AI. It really underscores
• 11:31
  that constant tension, doesn't it? Between pushing
• 11:34
  technology forward, ensuring our data, especially
• 11:37
  sensitive data, is protected and figuring out
• 11:40
  how we integrate increasingly powerful AI into
• 11:43
  everything. It's a balance we're constantly negotiating.
• 11:46
  What are the long -term consequences of these
• 11:48
  trends? Yeah. That's something worth thinking
• 11:50
  about. Definitely food for thought. And just
• 11:53
  a quick note, if the 23 Minute Story caught your
• 11:56
  interest and you want more background on their
• 11:58
  financial struggles and the bankruptcy itself.
• 12:00
  Right. We actually did a separate deeper dot
• 12:03
  session. Focus just on that. Exactly. You can
• 12:06
  find a link to that in our show notes if you
• 12:07
  want to dive deeper into the specifics of the
• 12:10
  23 and 8 bankruptcy. Great reminder. Well that
• 12:12
  wraps up our deep dive for today. Thanks for
• 12:14
  connecting the dots with us. Thanks everyone.
• 12:16
  See you next time. Thanks for listening, and
• 12:34
  we'll catch you next time.