Latest / Best of LinkedIn: Digital Transformation & Tech / Best of LinkedIn: ICT & Tech Insights CW 33/ 34

Transcript

Best of LinkedIn: ICT & Tech Insights CW 33/ 34 · Best of LinkedIn: Digital Transformation & Tech

• 0:00
  Speaker 1: This episode is provided by Thomas Allgaier and Frennus, based on the most relevant LinkedIn posts about ICT and tech insights from CW-III & XXXIV.
• 0:07
  Speaker 1: Frenness supports ICT enterprises in the form of delivering precise ICT market and pricing intelligence that analysts' subscriptions and existing databases cannot provide – you can find more info.
• 0:17
  Speaker 1: The security operations center at a major bank just lights up with this critical alert.
• 0:28
  Speaker 1: Oh boy!
• 0:29
  Speaker 1: Yeah, the privileged identity is in there accessing sensitive customer data it's rapidly firing off API calls trying to wire funds right and the security team is scrambling.
• 0:38
  Speaker 1: they're trying to figure out which employee A's account got compromised but thing is...there IS no employee.
• 0:43
  Speaker 1: Wow yeah..the identity actually belongs an AI agent that was officially authorized on the network.
• 0:49
  Speaker 0: That is the exact scenario that Ratan Jyoti mapped out recently, and I mean it's genuinely terrifying.
• 0:56
  Speaker 0: But anyway, welcome to Deep Dive.
• 0:58
  Speaker 0: Our mission today is highly focused.
• 0:59
  Speaker 0: we're analyzing the latest insights from top tech leaders to kind of separate The signal-from-the noise for ICD professionals Right and
• 1:05
  Speaker 1: that
• 1:06
  Speaker 0: two through three
• 1:06
  Speaker 1: day AM scenario you just mentioned.
• 1:08
  Speaker 1: That represents Just one Of Three Massive Interconnected Shifts We're Unpacking Today.
• 1:14
  Speaker 1: So we are moving From Traditional Cyber Compliance Into This Era Of AI Driven Resilience Tracking The Sudden Commercial Acceleration Of Quantum Computing.
• 1:23
  Speaker 1: And Then Finally Looking At The Hidden physical data infrastructure limits that are actively threatening all of this digital transformation.
• 1:31
  Speaker 0: It's a lot to cover, let's start with the first theme.
• 1:34
  Speaker 0: The timeline for cyber threats has basically fundamentally collapsed.
• 1:37
  Speaker 0: like we've always thought about security in terms of human reaction time right?
• 1:42
  Speaker 0: Getting an alert investigating it may be taking a server offline but humans just can't react fast enough.
• 1:47
  Speaker 0: if they attacker is algorithm They
• 1:49
  Speaker 1: absolutely cannot.
• 1:50
  Speaker 1: Philippe Sal pointed out attackers now breach network and pivot secondary systems under thirty minutes.
• 1:55
  Speaker 1: on average
• 1:56
  Speaker 0: Thirty minutes that's insane.
• 1:57
  Speaker 0: Yeah, if you think about like an old-school bank robbery the robbers are already inside The vault opening the safety deposit boxes before the front door alarm has even finished its first ring.
• 2:08
  Speaker 0: so AI detects threats at machine speed?
• 2:11
  Speaker 0: Sure, but a compromise AI agent with delegated authority is just a catastrophic attack surface.
• 2:18
  Speaker 0: Which
• 2:18
  Speaker 1: honestly makes sense of the massive industry update that Bob Carver flagged.
• 2:22
  Speaker 1: He highlighted open-AI leader Chris Lathain was warning about persistent AI driven attacks And threat was severe enough that they actually paused their advanced model development just due to safety fears around these offensive capabilities.
• 2:36
  Speaker 0: And we are seeing how those capabilities manifest in the wild.
• 2:40
  Speaker 0: Erin still well noted that autonomous agents are actively crossing real-world security boundaries, because we're letting AI systems act on our behalf across different platforms which just demands a total shift in how we isolate credentials.
• 2:53
  Speaker 1: It's sort of like giving a valet the keys to park your car but they technically hold the key.
• 2:57
  Speaker 1: so it decides to enter into an illegal street race.
• 3:01
  Speaker 1: If an A.I has the keys for your network...how do you even govern what it decides execute?
• 3:08
  Speaker 1: This completely breaks the traditional idea of zero trust, doesn't it?
• 3:11
  Speaker 1: Like if The AI is allowed to be there who do we blame when It goes rogue.
• 3:15
  Speaker 0: Is it an insider threat a privacy breach A failure of AI governance?
• 3:20
  Speaker 0: I mean the lines are completely blurred now.
• 3:22
  Speaker 1: Let me stop you There though because i kind Of want To challenge this narrative a bit.
• 3:26
  Speaker 1: If ai is this tireless infinitely scalable attacker.
• 3:30
  Speaker 1: Isn't it also the only viable defender we actually have left to fight back?
• 3:34
  Speaker 0: You've hit on the exact pivot happening in the industry right now.
• 3:37
  Speaker 0: Vasu Jekyll argues that agentic AI is what enables affordable, proactive defense for everyone.
• 3:43
  Speaker 0: It effectively turns security from a cost center into catalysts of innovation.
• 3:47
  Speaker 0: Interesting And Bipol Sinha put it very bluntly.
• 3:50
  Speaker 0: He said agentic cyber resilience isn't some future state...it's already here today.
• 3:54
  Speaker 1: Okay so the defensive tools exist But if the tools are AI-driven, how we test them has to evolve too.
• 4:01
  Speaker 1: Right?
• 4:01
  Speaker 1: Yeah.
• 4:01
  Speaker 1: Helen you made a point that we have to abandon plant and time compliance
• 4:05
  Speaker 0: right because an audit is just a snapshot.
• 4:07
  Speaker 0: being vulnerable on paper You know like having a known software bug doesn't necessarily mean an attacker can actually reach it to exploit It.
• 4:15
  Speaker 0: she argues for continuous Attacker centric testing To find the actual viable paths to your critical systems
• 4:22
  Speaker 1: which completely changes the conversation you have to have with The Boardroom.
• 4:27
  Speaker 0: Juan Pablo Castro noted that CISOs have to stop speaking in CVEs, those strings of numbers identifying specific software bugs and start speaking the actual language at risk.
• 4:38
  Speaker 1: Yeah, board doesn't care about a CVE number?
• 4:40
  Speaker 0: Not at all!
• 4:40
  Speaker 0: They care business trade-offs.
• 4:42
  Speaker 0: what consequence are we going to accept?
• 4:44
  Speaker 0: And just throwing money at the problem isn't always an answer either.
• 4:47
  Speaker 0: Shawnee Delaney highlighted that British intelligence ranked highest in security programs specifically because they proved threat.
• 4:53
  Speaker 0: literacy and continuous assessment beat massive software budgets,
• 4:57
  Speaker 1: especially when dealing with insider risk.
• 4:59
  Speaker 1: Exactly!
• 4:59
  Speaker 1: You can assess this posture right now.
• 5:01
  Speaker 1: Dr.
• 5:01
  Speaker 1: Erdal Oskia released a free twenty question resilience assessment That maps to five core NIST domains In literally six minutes.
• 5:10
  Speaker 1: Wow...that's fast
• 5:11
  Speaker 0: Right but actually implementing the frameworks takes deep operational change.
• 5:16
  Speaker 0: Thomas Roush shared that Hillshara's recent ISO twenty seven thousand one certification took months of embedding security into daily workflows.
• 5:25
  Speaker 0: It's not a checklist.
• 5:26
  Speaker 0: It is a culture shift.
• 5:28
  Speaker 0: Yeah, and for anyone on that journey Ellie Matar shared five free resources specifically for navigating ISO.
• 5:34
  Speaker 0: Twenty-seven thousand oh One.
• 5:36
  Speaker 1: but I mean none of those frameworks matter if your team doesn't understand the underlying plumbing Mm-hmm.
• 5:41
  Speaker 1: Mike Holcomb stressed a crucial point you have to learn networking before you touch OT or operational technology, cybersecurity.
• 5:49
  Speaker 1: You need to understand firewalls before you worry about quantum encryption and Jared Kuchit backed this up saying cyber security simply is not an entry-level field.
• 5:57
  Speaker 1: it's
• 5:57
  Speaker 0: just not
• 5:58
  Speaker 1: right.
• 5:58
  Speaker 1: requires a deep IT foundation.
• 6:00
  Speaker 1: yeah have to know how things like active directory in DNS actually work
• 6:05
  Speaker 0: because you need to know what normal looks like so that you can actually spot the abnormal.
• 6:10
  Speaker 0: And we are seeing vendors realize this need for foundational integration, too.
• 6:15
  Speaker 0: Brian Lannigan pointed out a partnership between Commvault and Google that embeds threat intelligence directly into recovery workflows... Oh!
• 6:22
  Speaker 0: That's small?
• 6:23
  Speaker 0: Yeah So if get hit You know exactly when data is clean before you restore it preventing secondary infection loop.
• 6:30
  Speaker 0: Marcel Velica mapped out fifteen combinations where isolated skills become massive force multipliers.
• 6:37
  Speaker 1: Like what?
• 6:37
  Speaker 0: For instance, combining threat intelligence with threat hunting or merging a sign which is basically a massive security camera system for your network With detection engineering Which is writing the rules of what those cameras should flag
• 6:50
  Speaker 1: And communicating.
• 6:51
  Speaker 1: all this complexity requires some serious storytelling.
• 6:54
  Speaker 1: Brittany Gannison introduced this brilliant concept called The Adventures Of Shep
• 6:58
  Speaker 0: Yeah.
• 6:59
  Speaker 1: It uses a great Pyrenees dog protecting the digital pasture to explain abstract concepts like zero trust, to non-technical employees.
• 7:07
  Speaker 0: Context is everything.
• 7:09
  Speaker 0: Sarah Fluss warned that while safety assessments and cybersecurity assessments need talk with each other they really must remain separate.
• 7:17
  Speaker 1: Why's it?
• 7:18
  Speaker 0: Well cyber risk about digital assets in data whereas safety risks are physical harm for humans.
• 7:24
  Speaker 0: The methodologies just fundamentally different.
• 7:26
  Speaker 1: That makes total sense.
• 7:28
  Speaker 1: And the stakes go way beyond individual companies too.
• 7:31
  Speaker 1: Chuck Brooks pointed out that AI and IoT are shifting our ultimate goal from protecting isolated networks to ensuring the viability of our entire digital ecosystem, and at NUMA it reminded us that cyber operations are now incredibly powerful foreign policy tools.
• 7:47
  Speaker 1: you can disrupt a nation's infrastructure without a single soldier ever crossing a physical border
• 7:52
  Speaker 0: And Rene Bonvane echoed that, noting how geopolitics actively reshapes cybersecurity priorities on a macro scale.
• 7:59
  Speaker 0: We're seeing massive capital deployment because of this.
• 8:02
  Speaker 0: Aramozal Omomhenle highlighted a one billion dollar alliance between NTT data and Palo Alto networks targeting these transformation areas, specifically autonomous operations an AI governance.
• 8:13
  Speaker 1: It's huge but underneath all that technology in Capital there is the human layer right?
• 8:18
  Speaker 1: Elid Franklin discussed the psychology resilience for on-call teams The people who actually carry the blame when things go down.
• 8:25
  Speaker 1: Yeah
• 8:26
  Speaker 0: they take the brunt
• 8:27
  Speaker 1: Absolutely.
• 8:28
  Speaker 1: And Dr Pat Billion expanded on that, explaining that leader burnout decision fatigue and lack of sleep are actively undermining enterprise resilience.
• 8:37
  Speaker 1: the human body is our ultimate.
• 8:39
  Speaker 1: an often most fragile infrastructure layer
• 8:42
  Speaker 0: in Human Fatigue Is a stark contrast to machines which are honestly only accelerating.
• 8:49
  Speaker 0: while AI reshapes today's security quantum computing is actively rewriting tomorrow's cryptography.
• 8:55
  Speaker 1: Yeah, before we jump into how quantum basically breaks the internet a quick reminder if you're enjoying this deep dive make sure to hit subscribe so don't miss our future editions.
• 9:03
  Speaker 1: Okay, Quantum!
• 9:05
  Speaker 1: The skepticism around it feels incredibly familiar doesn't
• 9:08
  Speaker 0: It does.
• 9:08
  Speaker 0: Leandro Corrente made a perfect comparison, he likened today's quantum skepticism to that famous nineteen ninety five interview where David Letterman mocked Bill Gates about the internet.
• 9:17
  Speaker 0: people say well don't traditional computers do just fine right
• 9:20
  Speaker 1: but they are entirely missing the exponential leap in capability
• 9:23
  Speaker 0: and The market isn't waiting for the skeptics to catch up either.
• 9:27
  Speaker 0: Isabel Busquette cited a BCG report showing That In twenty twenty-five enterprise Quantum spending actually overtook government And research lab spending.
• 9:37
  Speaker 0: Sixty-two percent of major surveyed firms are spending at least one million dollars a year on it.
• 9:42
  Speaker 0: Wow Philip in Talloura noted that HSBC is hiring a director of quantum for London and Singapore to build their strategy right now.
• 9:49
  Speaker 1: And they're hiring because the threat timeline is just shrinking.
• 9:53
  Speaker 1: Andrew Benigno built a model at the NSA predicting that quantum would break traditional encryption by late twenty-twenty eight, which aligns perfectly with IBM and Google publicly targeting twenty twenty nine for fault tolerant quantum
• 10:07
  Speaker 0: computing.
• 10:07
  Speaker 0: Let's unpack the math behind that threat for a second because David Steenhoek explained how Shor's algorithm actually makes this possible.
• 10:14
  Speaker 0: It is not just faster computer guessing passwords, it reduces factoring problem to finding period of modular exponential function.
• 10:21
  Speaker 1: What does that mean in plain English?
• 10:24
  Speaker 1: In plain english traditional computers try break encryption by brute forcing a lock guessing every key on massive ring.
• 10:33
  Speaker 1: Shor's algorithm uses quantum states to analyze the mechanical structure of a lock itself, finds the underlying mathematical pattern and just bypasses the pins entirely.
• 10:42
  Speaker 1: It runs in polynomial time completely shattering RSA encryption.
• 10:45
  Speaker 0: But here is the fascinating prist are we waiting purely on quantum hardware?
• 10:50
  Speaker 0: No Jaime Gomez Garcia noted that Anthropic's large language models actually found flaws in a post-quantum cryptographic candidate called HAWK, and it cost just one hundred thousand dollars in computing tokens.
• 11:03
  Speaker 0: You're
• 11:03
  Speaker 1: kidding?
• 11:04
  Speaker 0: Nope!
• 11:04
  Speaker 0: The AI models accelerated the discovery of mathematical weaknesses so efficiently that HAWk had to be withdrawn from standardization.
• 11:12
  Speaker 0: So AI is fundamentally changing the pace of cryptanalysis before quantum hardware even fully online
• 11:18
  Speaker 1: Which means we need crypto agility right now.
• 11:20
  Speaker 1: Rich Campania detailed how the industry is facing a cryptographic reset driven by ultra-short, forty seven days certificate life cycles and these harvest now decrypt later attacks.
• 11:29
  Speaker 0: Yeah those are scary.
• 11:30
  Speaker 1: Very.
• 11:31
  Speaker 1: That's where adversaries steal your encrypted data today hoard it And just wait for a quantum computer to read in five years.
• 11:38
  Speaker 1: Teams have been forced use tools like PANOS twelve point two To automate Certificate Life Cycles because humans can't keep up
• 11:46
  Speaker 0: And governments are stepping in.
• 11:47
  Speaker 0: Nicholas Friedman highlighted the proposed Quantum Guard Act, which pushes electric utilities to inventory their cryptography across all IT and OT environments.
• 11:56
  Speaker 0: You simply cannot migrate a network to quantum safe standards if you haven't mapped where the old encryption
• 12:01
  Speaker 0: lives."
• 12:02
  Speaker 1: And Robert Rogowski pointed out this applies to decentralized networks too... And we are literally testing this in extreme environments now.
• 12:16
  Speaker 1: Matthew R. shared that IBM tested quantum-safe cryptography and space, like on the International Space Station to secure satellite
• 12:23
  Speaker 1: infrastructure."
• 12:23
  Speaker 0: The hardware leaps making all of these necessary are staggering!
• 12:27
  Speaker 0: Pat Gelsinger noted IBM's trusted milestones are delivering results scientists can actually build upon — for example Ken Kwong and Dr.
• 12:35
  Speaker 0: Cory O'Mara discussed a massive step toward IBMs'.
• 12:40
  Speaker 0: They successfully linked and cooled modular cryogenic systems.
• 12:45
  Speaker 1: And how big are those?
• 12:46
  Speaker 0: We're talking about units the size of three kitchen fridges, holding three cubic meters vacuum volume reaching four kelvin which is colder than deep space in under five days.
• 12:55
  Speaker 1: That's wild!
• 12:57
  Speaker 1: The performance has already materializing.
• 13:00
  Speaker 1: Jan McAllen shared that IBM hit quantum advantage solving a specific problem within fifteen minutes that would completely overwhelm classical supercomputers.
• 13:09
  Speaker 0: Because of that, the broader ecosystem is mobilizing really fast.
• 13:13
  Speaker 0: Greg Karman emphasized that commercializing deep tech means preparing your business use cases.
• 13:17
  Speaker 0: today.
• 13:18
  Speaker 0: Ravi Gupta is evaluating structured methods for applying quantum to molecular discovery for pharmaceuticals.
• 13:23
  Speaker 0: Huge
• 13:24
  Speaker 1: potential there!
• 13:25
  Speaker 0: Yeah.
• 13:25
  Speaker 0: and Nathan Thomas talked about bringing Helios Quantum Hardware directly to Oracle Cloud so developers don't have to build specialized highly chilled infrastructure just to experiment
• 13:35
  Speaker 1: The goal.
• 13:38
  Speaker 1: Shrikantu Udumu published a paper on making intent-level quantum programming accessible to students, and the infrastructure investments are just global at this point.
• 13:47
  Speaker 1: Srinivasa Rauluri detailed Canada's twenty point three million dollar Quantum Defense Innovation Secure Hub to test next gen sensing.
• 13:54
  Speaker 0: Nice!
• 13:55
  Speaker 1: And Becky Carroll clarified that The New Illinois Quantum Park is dedicated innovation campus relying on specialized helium cooling.
• 14:03
  Speaker 1: it's vastly different from water heavy classical data center.
• 14:06
  Speaker 0: John Head is writing about that exact massive quantum opportunity emerging in Chicago.
• 14:11
  Speaker 0: While globally, Sanjeev Singh highlighted QPAI building a full-stack quantum ecosystem over India.
• 14:17
  Speaker 0: What
• 14:17
  Speaker 1: I find mindblowing is how the physical architecture might evolve.
• 14:20
  Speaker 1: Roderto Siagri suggested quantum won't just be massive centralized mainframes forever.
• 14:26
  Speaker 1: it might actually follow the path of GPUs evolving into small distributed accelerators integrated directly to standard edge servers.
• 14:34
  Speaker 0: Marco Antonio D'Anessiamento took the security implications even further.
• 14:38
  Speaker 0: He argued that in a true quantum environment, all static encryption is effectively transparent.
• 14:44
  Speaker 0: Transparent?
• 14:45
  Speaker 0: Yeah.
• 14:45
  Speaker 0: so he says security must shift entirely to hardware wing zero!
• 14:50
  Speaker 0: That's the absolute... deepest, most privileged level of a computer processor utilizing wetware biochips.
• 14:56
  Speaker 0: Wait!
• 14:56
  Speaker 1: Wetware?
• 14:57
  Speaker 1: Yes
• 14:58
  Speaker 0: we're talking about blending biological materials with silicon to enforce state immutability at the physical level basically proving data hasn't been tampered with.
• 15:06
  Speaker 1: Okay that sounds like a sci-fi movie but business models around this deep tech are maturing too right?
• 15:11
  Speaker 1: Jan Goetz outlined IQMera's dual NASDAQ and Helsinki listing as blueprint for European Tech IPOs.
• 15:18
  Speaker 1: That combines Deep Technical Roots With US Capital Though
• 15:20
  Speaker 0: it is not all just serious boardrooms and biochips.
• 15:24
  Speaker 0: Karel Piscan joked about inventing the Garris-Beckstein constant during a four hundred kilometer cycling tour across Swiss quantum hubs.
• 15:32
  Speaker 1: What's The Constant?
• 15:33
  Speaker 0: It represents the agonizing two kilometers that magically always seem to remain before you finally reach your destination.
• 15:40
  Speaker 1: I think every cyclist has felt that constant.
• 15:42
  Speaker 1: But this brings us to our final theme, which is honestly the ultimate reality check.
• 15:47
  Speaker 1: Neither agentic AI nor quantum computing matters if your basic data infrastructure and digital plumbing fail.
• 15:55
  Speaker 1: We need places to actually plug these things in!
• 15:57
  Speaker 0: Exactly, Archie Coffield shared a prime example.
• 16:00
  Speaker 0: AWS is planning an eighteen building three point nine six million square foot data center campus in Ohio.
• 16:07
  Speaker 0: The footprint is massive, but space water and glid power are actively running out.
• 16:12
  Speaker 0: We're hitting the physical limits of the electrical grid
• 16:15
  Speaker 1: which makes what Brendan Wallace argued sound a lot less crazy.
• 16:18
  Speaker 1: He pointed to Space Base data centers.
• 16:20
  Speaker 1: right sounds like science fiction But he argues they were legitimate inevitable real estate response To terrestrial supply constraints.
• 16:28
  Speaker 0: And for the infrastructure that remains on earth?
• 16:30
  Speaker 0: Physical resilience Is completely non-negotiable.
• 16:33
  Speaker 0: Gunis and Dharan mapped out the journey from N plus one to two end fault tolerant architecture.
• 16:40
  Speaker 1: Let me jump in with an analogy for that real quick go for it.
• 16:42
  Speaker 1: So n-plus.
• 16:43
  Speaker 1: One is kind of like having a spare tire on your trunk.
• 16:45
  Speaker 1: if one thing fails you have a backup component But two end is like having an entire second car driving next to you, but its own engine and fuel.
• 16:56
  Speaker 1: It means independent power pads dual carriers, redundant chillers.
• 17:00
  Speaker 1: A single failure somewhere in the facility simply cannot cause a business
• 17:03
  Speaker 1: outage.".
• 17:04
  Speaker 0: And Patrick Osborne noted that this physical reliance ties into The Great VM Reset where organizations are completely rethinking the economics of their virtualization software while simultaneously expanding to sovereign cloud and heavy AI workloads...
• 17:17
  Speaker 1: But hardware and data centers really only have the battle Manuel Berrigan warned about the software layer If your employees are acting as manual data bridges, literally copy pasting information from one platform to another because the systems don't talk in.
• 17:30
  Speaker 1: Your tech stack is fundamentally broken.
• 17:32
  Speaker 1: you Don't have a people problem.
• 17:33
  Speaker 1: You Have A Data Integration Problem?
• 17:36
  Speaker 1: Your Employees Are Functioning As Human APIs
• 17:40
  Speaker 0: And This Exposes The Hidden Costs Of Transformation.
• 17:43
  Speaker 0: Jeff Winter Called Technology Pilots A Trojan Horse.
• 17:47
  Speaker 0: The initial software pilot isn't the expensive part.
• 17:50
  Speaker 0: The cost is dozens of integrations, massive data cleanups and ongoing support models hiding inside that pilot.
• 17:57
  Speaker 1: George Collier took it even further.
• 17:59
  Speaker 1: He stated ERP transformations overhauling a company's core software rarely lose money because technology itself is bad.
• 18:07
  Speaker 1: They lose money on poor early decisions Things like picking up a software vendor before understanding root business problems.
• 18:14
  Speaker 1: We're spending millions to customize a brand new system just to preserve ten years of bad legacy processes.
• 18:19
  Speaker 0: And this extends far beyond enterprise headaches, This affects national economies.
• 18:24
  Speaker 0: Confidence Daveley highlighted Nigeria's New National Digital Cloud Policy which promises A seven hundred and fifty million dollar investment over twenty four months.
• 18:33
  Speaker 0: That
• 18:33
  Speaker 1: is the huge commitment
• 18:34
  Speaker 0: It Is.
• 18:35
  Speaker 0: But Lucia Baji accurately noted that digital transformation in a country like Nigeria only creates value if it solves real governance challenges enabling seamless data exchange to actually deliver better public services.
• 18:49
  Speaker 1: Ashley DeBebe proposed a fascinating classification framework for Ethiopia to navigate this, it helps the government decide which critical software should be built in-house by Public Agencies For Security versus what should be competitively delivered by private tech firms
• 19:04
  Speaker 0: because implementation at any scale is incredibly messy.
• 19:08
  Speaker 0: Take AI integration, for example.
• 19:10
  Speaker 0: Muhammad Syed pointed out that when rag co-pilots retrieval augmented generation hallucinate and give bad answers teams usually blame the embedding models.
• 19:18
  Speaker 1: Let's unpack Argi really quickly.
• 19:20
  Speaker 1: think of Argi like giving an AI in open book test.
• 19:23
  Speaker 1: Okay instead of relying on his memory it searches your company's documents for the answer.
• 19:27
  Speaker 1: Syed is saying.
• 19:28
  Speaker 1: The problem isn't the AIs memory.
• 19:30
  Speaker 1: the problem is shallow retrieval depth And a missing re ranking step.
• 19:34
  Speaker 1: Basically, the AI pulls a paragraph based on keyword similarity but fails to verify if that paragraphs actually answers the intent of the user's question.
• 19:42
  Speaker 0: So it is just guessing poorly?
• 19:44
  Speaker 0: Right!
• 19:44
  Speaker 1: If you skip that re-ranking quality check your shiny new compliance bot will confidently cite the absolute wrong regulation.
• 19:52
  Speaker 0: And that nuance is exactly why deep engineering integration matters so much.
• 19:57
  Speaker 0: Colleen Capais highlighted that OpenAI had to partner with IBM, to launch a dedicated practice simply because modernizing complex enterprise workflows requires massive specialized engineering expertise.
• 20:09
  Speaker 0: You can't just plug it in.
• 20:11
  Speaker 0: Vishakasadwani praised Akamai's developer hub for actually showing the messy trade-offs and limits of building on
• 20:30
  Speaker 1: all the way up to complex edge computing.
• 20:32
  Speaker 1: It's
• 20:32
  Speaker 0: intense
• 20:33
  Speaker 1: and Ramel Marin announced a global distribution of Flapsat v one point zero, which allows the aerospace hacker community to experiment with satellite hardware locally just getting their hands dirty with the actual physical limits of the tech.
• 20:45
  Speaker 0: it All comes back to a foundational mindset.
• 20:48
  Speaker 0: Christopher Jones said resilient connectivity like super broadband is The bedrock that enables all of this adaptability productivity And AI adoption.
• 20:57
  Speaker 0: if the pipes are broken nothing flows.
• 21:00
  Speaker 1: And Bob Fabian Zenga, in his review of the book The Kill Chain reminded us of the ultimate truth tying all three themes together today.
• 21:07
  Speaker 1: Future conflicts and enterprise survival won't be one simply by purchasing superior technology.
• 21:12
  Speaker 1: they will be through adaptive leadership and building networked resilient systems that can sense an act faster than disruption around them.
• 21:21
  Speaker 0: The real differentiator is how quickly an organization can reconfigure itself when the ground shifts.
• 21:28
  Speaker 0: Which leaves us with this final thought, we are obsessing over processing speed of AI agents and cryogenic temperatures on quantum hardware.
• 21:35
  Speaker 0: but what if a real bottleneck isn't silicon?
• 21:38
  Speaker 0: What if it's human adaptability?
• 21:41
  Speaker 0: If technology evolves exponentially will our legacy organizational structure survive or rewrite what a company looks like?
• 21:52
  Speaker 1: Also, check out our other editions on Cloud Insights and Sovereignty, Digital Products & Services, AI and Agenetic Systems, Green ICT and Sustainable AI, DefenseTech and HealthTech.