Quote for the day:
"The only way to achieve the impossible is to believe it is possible." -- Charles Kingsleigh
Way too complex: why modern tech stacks need observability
Recent outages have demonstrated that a heavy dependence on digital systems
can leading to cascading faults that can halt financial transactions, disrupt
public transportation and even bring airport operations to a standstill. ...
To operate with confidence, businesses must see across their entire digital
supply chain, which is not possible with basic monitoring. Unlike traditional
monitoring, which often focuses on siloed metrics or alerts, observability
provides a unified, real-time view across the entire technology stack,
enabling faster, data-driven decisions at scale. Implementing real-time,
AI-powered observability covers every component from infrastructure and
services to applications and user experience. ... Observability also enables
organizations to proactively detect anomalies before they escalate into
outages, quickly pinpoint root causes across complex, distributed systems and
automate response actions to reduce mean time to resolution (MTTR). The result
is faster, smarter and more resilient operations, giving teams the confidence
to innovate without compromising system stability, a critical advantage in a
world where digital resilience and speed must go hand in hand. Resilient
systems must absorb shocks without breaking. This requires both cultural and
technical investment, from embracing shared accountability across teams to
adopting modern deployment strategies like canary releases, blue/green
rollouts and feature flagging. Radical Empowerment From Your Leadership: Understood by Few, Essential for All
“Radical empowerment, for me, isn’t about handing people a seat at the table.
It’s about making sure they know the seat is already theirs,” said Trenika
Fields, Business Legal, AI Leader at Cisco, MIT Sloan EMBA Class of ’26. “I
set the vision and I trust my team to execute in ways that are anchored in the
mission and tied to real business outcomes. But trust without depth doesn’t
work. That’s where leading with empathy comes in. It’s my secret sauce, and it
has to be real. You can’t fake it. People know when it’s performative. Real
empathy builds confidence, and confidence fuels bold, decisive execution. When
people feel seen, trusted, and strategically aligned, they lead like builders,
not bystanders. Strip that trust and empathy away, and radical disempowerment
moves in fast. Voices go quiet. Momentum dies. Innovation flatlines. But when
you get it right, you don’t just build teams. You build powerhouses that set
the standard and raise the bar for everyone else.” Why, given how simple this
is, is it so hard for senior leadership to do versus say? I worked in an
environment years ago when “radical candor” was the theme du jour rather than
“radical empowerment.” An executive over an executive over my boss was
explaining radical candor, which very simply put, being constructive and
forthright with empathy to help others grow. Banks Can Convert Messy Data into Unstoppable Growth
Banks recognize the potential in tapping a trove of customer data, much of it
unstructured, as a tool to personalize interactions and become more proactive.
They are sitting on a goldmine of unstructured information hidden in PDFs,
scanned forms, call notes and emails — data that, once cleaned and organized,
can unlock new business opportunities, says Drew Singer, head of product at
Middesk. ... The ability to successfully turn data into insights often depends
on clear parameters for how data is handled. This includes a shared
understanding of who owns the data, how it will be managed and stored, and a
defined governance structure — possibly through committees — for overseeing
its use, Deutsch says. "If you don’t set these rules, once data starts
flowing, you will lose control of it. You will most likely lose quality," he
says. ... With the data governance structure firmly in place, FIs are
positioned to use additional tools to garner action-oriented insights across
the organization. Truist Client Pulse, for example, uses AI and machine
learning to analyze customer feedback across channels. ... "We’ve got a
population of teammates using the tool as it stands today, to better
understand regional performance opportunities …what’s going well with certain
solutions that we have, and where there are areas of opportunity to enhance
experience and elevate satisfaction to drive to client loyalty," says
Graziano. Securing Digital Supply Chains: Confronting Cyber Threats in Logistics Networks
Modern logistics networks are filled with connected devices — from IoT sensors
tracking shipments and telematics in trucks, to automated sorting systems and
industrial controls in smart warehouses and ports. This Internet of Things
(IoT) revolution offers incredible efficiency and real-time visibility, but it
also increases the attack surface. Each connected sensor, RFID reader, camera,
or vehicle telemetry unit is essentially an internet entry point that could be
exploited if not properly secured. The spread of IoT devices introduces new
vulnerabilities that must be managed effectively. For example, a hacker who
hijacks a vulnerable warehouse camera or temperature sensor might find a way
into the larger corporate network. ... The tightly interwoven nature of modern
supply chains amplifies the impact of any single cyber incident, highlighting
the importance of robust cybersecurity measures. Companies are now digitally
linked with vendors and logistics partners, sharing data and connecting
systems to improve efficiency. However, this interdependence means that a
security failure at one point can quickly spread outward. ... While large
enterprises may invest heavily in cybersecurity, they often depend on smaller
partners who might lack the same resources or maturity. Global supply chains
can involve hundreds of suppliers and service providers with varying security
levels.
For OT Cyber Defenders, Lack of Data Is the Biggest Threat
Data in the OT and ICS world is transient, said Lee. Instructions - legitimate,
or not - flow across the network. Once executed, they vanish. "If I don't
capture it during the attack, it's gone," Lee said. Post-incident forensics is
basically impossible without specialized monitoring tools already in place. "So
for the companies that aren't doing that data collection, that monitoring, prior
to the attacks, they have no chance at actually figuring out if a cyberattack
was involved or not." And that is a problem when nation-state adversaries have
pre-positioned themselves within the networks of critical infrastructure
providers, apparently ready to pivot to OT exploitation in time of conflict. ...
Even when critical infrastructure operators do capture OT monitoring data, the
sheer complexity of modern industrial processes means that finding out what went
wrong is difficult. The inability to make use of more detailed data is an
indicator of immaturity in the OT security space, Bryson Bort told Information
Security Media Group. "The way I summarize the OT space is, it's a generation
behind traditional IT," said Bort, a U.S. Army veteran and founder of the
non-profit ICS Village. Bort helps organize the annual Hack the Capitol event,
but he makes his living selling security services to critical infrastructure
owners and operators. Most operators still don't have visibility into the ICS
devices on their work, Bort said. "What do I have? What assets are on my
network?"
Cross-Border Compliance: Navigating Multi-Jurisdictional Risk with AI
The digital age has turned global expansion from an aspiration into a necessity.
Yet, for companies operating across multiple countries, this opportunity comes
wrapped in a Gordian knot of cross-border compliance. The sheer volume,
complexity, and rapid change of multi-jurisdictional regulations—from GDPR and
CCPA on data privacy to complex Anti-Money Laundering (AML) and financial
reporting rules—pose an existential risk. What seems like a local detail in one
jurisdiction may spiral into a costly mistake elsewhere. ... AI helps with
cross-border compliance by automating risk management through real-time
monitoring, analyzing vast datasets to detect fraud, and keeping up with
constantly changing regulations. It navigates complex rules by using natural
language processing (NLP) to interpret regulatory texts and automating tasks
like document verification for KYC/KYB processes. By providing continuous,
automated risk assessments and streamlining compliance workflows, AI reduces
human error, improves efficiency, and ensures ongoing adherence to global
requirements. AI, specifically through technologies like Machine Learning (ML)
and Natural Language Processing (NLP), is the critical tool for cutting
compliance costs by up to 50% while drastically improving accuracy and speed. AI
and machine learning (ML) solutions, often referred to as RegTech, are
streamlining compliance by automating tasks, enhancing data analysis, and
providing real-time insights.
Best Practices for Building an AI-Powered OT Cybersecurity Strategy
One challenge in defending OT assets is that most industrial facilities still
rely on decades-old hardware and software systems that were not designed with
modern cybersecurity in mind. These legacy systems are often difficult to patch
and contain documented vulnerabilities. Sophisticated adversaries know this and
exploit these outdated systems as a point of entry. ... OT cybersecurity and
regulatory compliance are tightly linked in manufacturing, but not
interchangeable. Consider regulatory compliance the minimum bar you must clear
to stay legally and contractually safe. At the same time, cybersecurity is the
continuous effort you must take to protect your systems and operations.
Manufacturers increasingly must prove OT cyber resilience to customers,
partners, and regulators. A strong cybersecurity posture helps ensure
certifications are passed, contracts are won, and reputations are protected. ...
AI is a powerful tool for bolstering OT cybersecurity strategies by overcoming
the common limitations of traditional, rule-based defenses. AI, whether machine
learning, predictive AI, or agentic AI, provides advanced capabilities to help
defenders detect threats, automate responses, manage assets, and enhance
vulnerability management. ... Human oversight and expertise are vital for
ensuring AI quality and contextual accuracy, especially in safety-critical OT
environments. Training Data Preprocessing for Text-to-Video Models
/articles/training-data-preprocessing-for-text-to-video-models/en/smallimage/training-data-preprocessing-for-text-to-video-models-thumbnail-1762244675850.jpg)
Getting videos ready for a dataset is not merely a checkbox task - it’s a
demanding, time-consuming process that can make or break the final model. At
this stage, you’re typically dealing with a large collection of raw footage with
no labels, no descriptions, and at best limited metadata like resolution or
duration. If the sourcing process was well-structured, you might have videos
grouped by domain or category, but even then, they’re not ready for training.
The problems are straightforward but critical: there’s no guiding information
(captions or prompts) for the model to learn from, and the clips are often far
too long for most generative architectures, which tend to work with a context
window (length of the video, like number of tokens for Large Language Models)
measured in tens of seconds, not minutes. ... It might seem like the fastest
approach is to label every scene you have. In reality, that’s a direct route to
poor results. After all the previous steps, a dataset is rarely clean: it almost
always contains broken clips, low-quality frames, and clusters of near-identical
segments. The filtering stage exists to strip out this noise, leaving the
model only with content worth learning from. This ensures that the model doesn’t
spend time on data that won’t improve its output. ... Building a proper
text-to-video dataset is an extremely complex task. However, it is impossible to
build a text-to-video generation model without a good dataset.
Putting Design Thinking into Practice: A Step-by-Step Guide
The key aim of this part of the design process is to frame your problem
statement. This will guide the rest of your process. Once you’ve gathered
insights from your users, the next step is to distil everything down to the real
issue. There are many ways to do this, but if you’ve spoken to several users,
start by analysing what they said to find patterns — what themes keep coming up,
and what challenges do they all seem to face? ... Once you’ve got your problem
statement, the next step is to start coming up with ideas. This is the fun part!
The aim of this part of idea generation is not to find the perfect idea straight
away, but to come up with as many ideas as possible. Quantity matters more than
quality right now. Start by brainstorming everything that comes to mind, no
matter how unrealistic it sounds. At this point, quantity matters more than
quality — you can always refine later. Write your ideas down, sketch them, or
talk them through with friends or teammates. You might be surprised at how one
silly suggestion sparks a genuinely good idea. ... Testing is the “last” stage
of the design process. I say last with a bit of hesitation, because while it is
technically last on the diagram, you are guaranteed to get a lot of feedback
that will require you to go back to earlier stages of the design process and
revisit ideas.Beyond Resilience: How AI and Digital Twin technology are rewriting the rules of supply chain recovery
For decades, supply chain resilience meant having backup plans, alternate
suppliers, safety stock, and crisis playbooks. That model doesn’t hold anymore.
In a post-pandemic world shaped by trade wars, climate volatility, and
technology shocks, disruptions are neither rare nor isolated. They’re
structural. ... The KPIs of resilience have evolved. In most companies,
traditional metrics like on-time delivery or supplier lead time fail to capture
the system’s true flexibility. Modern analytics teams are redefining the
measurement architecture around three key indicators: Mean time to recovery
(MTTR): the time between initial disruption and full operational
stability; Conditional value-at-risk (CVaR): a probabilistic measure of
financial exposure under extreme stress; Supply network resilience index (SNRI):
a composite score tracking substitution agility and cross-tier visibility. ... A
hidden benefit of this new approach is its environmental alignment. When
Schneider Electric built a multi-tier AI twin for its Asia-Pacific operations,
it discovered that optimizing for resilience, diversifying ports, balancing lead
times, and automating inventory allocation also reduced carbon intensity per
unit shipped by 12%; This was not the goal, but it proved that sustainability
and resilience share a common denominator: Efficiency. The smarter the network,
the smaller its waste footprint. In boardrooms today, that realization is
quietly rewriting ESG strategy.
No comments:
Post a Comment