The massive ecosystem of IoT Edge devices brings along with it the most diverse sources of data, delivered across modalities ranging from text, vision, and voice to time series and touch. These data streams now power AI-accelerated device functions such as person detection, speech recognition, voice biometrics, adaptive noise reduction, echo cancellation, machine translation, text summarization, robotic gripping and a host of other multimodal applications. A transformation at this scale can only be enabled through advances in Edge AI silicon architecture and associated software – a combination necessary to unlock ambient intelligence at or near the source of sensor data that extend cloud-based experiences to devices, while delivering privacy and latency benefits.

More efficient AI models and open development tools are helping developers move from concept to deployment faster, turning on-device inference from an interesting possibility into a practical development path.

Edge AI is Enabling a New Innovation Frontier

Conversations about Edge AI often begin with three key considerations: latency, data privacy, and cloud costs. Bringing intelligence to the Edge can reduce round trips to the cloud, improve responsiveness, enhance privacy, and lower dependence on upstream bandwidth. But the bigger opportunity lies in what those advantages make possible.

A device that can see, hear, interpret, and respond locally is not just faster. It is more independent. It can keep functioning when connectivity is poor, reduce the need to send sensitive raw data to the cloud, and deliver experiences that feel natural because decisions are made closer to where the data is created.

Always-On Experiences

Edge device categories like wearables, hearables, smart appliances, automation hubs, and industrial endpoints are often constrained by battery life, thermal limits, size, or some combination of these factors. The right tools can lower the barrier to building ultra-low-power, always-on applications that use ambient sensing to deliver richer AI-enabled experiences that are practical, resilient, and sustainable in daily use.

Product teams need confidence that the systems they build today can adapt and remain relevant even as AI models evolve. The right platform should make experimentation less risky, not more.

Multimodal Systems

Consider what becomes possible when intelligence moves closer to where data is created:

• A smart home hub can process audio and vision locally to detect events and trigger responses.
• A wearable or hearable device can combine voice, motion, and environmental awareness while reducing cloud dependence.
• An industrial cobot can process camera feeds and other sensor data locally so it can respond quickly to activity in a workspace shared with people.

For Edge AI applications to reach their full potential, developers need platforms that support multimodal processing across vision, audio, sensor, and language data, allowing devices to interpret real-world context more fully. Moving from isolated Edge inference to integrated systems that combine compute, connectivity, sensing, and AI into cohesive, intelligent platforms is a shift that will expand what developers can build.

Creating a System that Reduces Edge AI Developer Friction

Fragmented ecosystems, closed toolchains, and difficulty integrating advanced AI into resource-constrained devices can all prevent developers from taking advantage of on-device AI. As AI models, optimization techniques, and frameworks evolve quickly, proprietary stacks can slow broad adoption, creating too much distance between a promising prototype and a production-ready system. If Edge AI remains hard to access, hard to optimize, or locked into fragmented workflows, only a narrow set of teams can use it well.

That is why the real opportunity is not just more capable Edge AI hardware. It is about coming together to establish an ecosystem built on open, standards-based, developer-friendly hardware, software and tooling. Reference platforms with native support for lightweight open models make Edge AI more accessible, lowering barriers to on-device AI innovation and accelerating time to product.

This is the philosophy behind Google Research and Synaptics’ collaboration that has resulted in the Synaptics Coralboard™: a development board designed to enable a new class of intelligent, always-on Edge AI applications where performance, power efficiency, and on-device intelligence define the user experience. Powered by the Synaptics Astra® SL2619 processor, the Coralboard combines a 1 TOPS Torq™ NPU with the industry’s first implementation of the Coral NPU™ from Google Research. Based on a RISC-V architecture, the Coral NPU presents a lightweight, C-programmable frontend designed to handle compute tasks and control flows. Put together, it provides a compact, standards-based developer-ready platform with the interfaces needed for real-world prototyping. The MLIR-based Torq open-source toolchain supports popular machine learning frameworks and models, providing a unified path from experimentation and optimization to deployment. When coupled with native support for Gemma, Google’s family of open models for the Edge, the combined hardware and software stack offers a foundation for building private, efficient and multimodal Edge AI applications.

As AI capabilities expand across everyday devices, the Synaptics Coralboard gives developers a quicker way to build production-grade Edge AI solutions efficiently.

Seizing the Potential of Edge AI

The Edge can become both an extension of experiences that originate in the cloud, as well as a source of enriched ambient context-aware intelligence, securely delivered through low-power devices. With the right mix of open models and open-source tools, scalable future-ready NPU architectures, and developer friendly platforms such as the Coralboard, more teams can start building innovative multimodal experiences on-device. It is a chance to build Edge AI systems that are efficient and private, and more human in the way they respond to the world.

The real opportunity for developers is about having the right toolkits, and true freedom to extend intelligence to where it works best.

When we talk about cybersecurity, we often default to averages.

The average number of threats blocked.
The average number of attacks per household.
The average level of risk.

Averages help us understand the scale of a problem. They’re much less useful for understanding where action is needed.

That’s one of the biggest takeaways from my latest byline in The Fast Mode, where I explore new research from Plume showing that a remarkably small percentage of connected homes account for most cybersecurity threat activity. Rather than being evenly distributed, risk is highly concentrated, with a tiny fraction of households experiencing persistent, repeated attacks. That finding has implications far beyond cybersecurity.

Read the full article on The Fast Mode

For years, broadband providers have relied on aggregate metrics to understand network health and subscriber experience.

But connected homes don’t behave like averages.

Some households rarely encounter security threats. Others experience them every day. Treating those homes as statistically equivalent masks the very subscribers who need the most protection.

This isn’t unique to security. It’s increasingly true across the connected home.

Some households stream occasionally. Others run multiple simultaneous video calls while gaming and interacting with AI assistants. Some homes have a handful of devices. Others have dozens.

The challenge for providers isn’t understanding the “average” subscriber; it’s recognizing that every connected home is different.

This is where AI and global intelligence become essential.

Identifying concentrated patterns of risk requires more than visibility into a single network. It requires learning across millions of homes, hundreds of operators, and billions of real-world interactions to distinguish isolated events from meaningful trends.

At Plume, our platform continuously learns across more than 600 million connected devices, over 50 million homes, 450+ ISP partners, and 58 countries. That global perspective allows us to identify patterns that would remain invisible within any individual network.

One example is the concentration of cybersecurity threats.

Tomorrow, those same capabilities will help providers anticipate entirely different kinds of subscriber challenges, from AI-driven application behavior to emerging device ecosystems and new network demands.

Perhaps the most important implication of this research is that cybersecurity shouldn’t simply be measured by the number of threats blocked.

Its value lies in protecting subscribers before they ever realize they’re under attack.

The same philosophy increasingly applies across the connected home.

Subscribers don’t think about packet loss, roaming algorithms, or DNS filtering. They simply expect their connected home to work reliably and securely.

That’s why the conversation is shifting from reacting to incidents toward proactively delivering better outcomes.

The providers that can identify problems early, personalize protection, and continuously optimize each subscriber’s experience won’t just operate more efficiently. They’ll build something far more valuable: confidence.

One statistic from the research still stands out: A tiny fraction of homes drives most observed threat activity.

It’s a reminder that the most important insights often aren’t found in the average. They’re found in the outliers.

As AI continues to reshape broadband, providers will need to lean into it to move beyond one-size-fits-all approaches and deliver experiences tailored to the realities of each connected home.

That begins with understanding where risk truly exists.

For a deeper look at the research and what it means for broadband providers, you can read my full article in The Fast Mode.

Read it here: A Tiny Fraction of Homes Drives the Majority of Threat Volume

If you’ve been following along with our 100G Ethernet series, you’ve run into a lot of vocabulary that gets used correctly but rarely explained: NRZ, PAM4, insertion loss, Type A versus Type B polarity, and a half-dozen acronyms that all seem to start with a letter that stands for “fiber.” This post is the answer key.

Nothing here is exotic. Most of it is old physics and older standards wearing new urgency, because the same conditions that a 10G link shrugged off for a decade will take down a 25G lane without warning. Bookmark this one — we’ll link back to it instead of re-explaining the same term in every post.

ASIC (Switch ASIC)
The dedicated chip inside a switch that actually moves traffic. Its design and the module you plug in decide which combinations of port speeds a switch can run at once.

BER (Bit Error Rate)
What fraction of bits arrive wrong, written in scientific shorthand: 1e-12 is one bad bit per trillion and healthy; 4e-5 is one per 25,000 and a lane about to become a ticket.

Cassette, Trunk, Patch Cord
The three pieces of a fiber channel. A trunk is the permanent multi-fiber cable running between rooms; a cassette is the panel-mounted box that breaks the trunk out into connectors; a patch cord is the short jumper from panel to equipment. Polarity has to be right across all three, and swapping one can break a channel that worked yesterday.

Clause 82 / Clause 91
Two sections of the IEEE 802.3 standard: one covers how a 100G stream is encoded and split across lanes, the other covers how its errors are corrected. You don’t need the internals — the operationally relevant fact is that on links using this encoding, FEC is a setting a human turns on at each end, and nothing in the protocol reconciles the two ends if they disagree.

CRC / FCS Errors and Discards
Three things your switch counts, and none of them mean quite what they sound like at 100G. A CRC error — which is what most switch CLIs print, though the more precise term is FCS (Frame Check Sequence), and an FCS is a type of CRC — means bit errors exceeded what FEC could correct, and the frame arrived bad. A discard means a frame was dropped on purpose, usually for congestion. FEC repairs damage before either counter is ever checked, so a lane can be bleeding errors quietly while both read zero.

CWDM (Coarse Wavelength Division Multiplexing)
Putting several signals on one fiber by giving each its own color of light. “Coarse” means the colors are spaced far enough apart to keep the optics cheap.

DAC (Direct Attach Copper)
A copper cable with a transceiver permanently attached at each end — no separate optic, no fiber required. Cheap and low-latency over short runs (typically a few meters), but heavier and less flexible than fiber, which is why you’ll find it inside a rack and rarely between racks.

FEC and RS-FEC
Forward Error Correction is extra data sent alongside the real data so the receiver can repair damaged bits itself, with no retransmission. RS-FEC is the specific Reed-Solomon scheme used at 25G and 100G, and it’s strong enough that a badly degraded lane gets repaired completely — which is exactly why the switch’s counters can read zero on a link that’s actually in trouble.

HCL (Hardware Compatibility List)
Your vendor’s official list of exactly which modules, cables, and configurations are supported together. Skipping it doesn’t always break something — but when it does, you’ll be troubleshooting alone instead of with a support contract behind you.

Insertion Loss
How much signal strength is lost end to end across a channel, measured in decibels. Every connector and splice costs a little; the question is whether the total stays inside the budget for the speed you’re running.

Margin
How much abuse a link can absorb and still work — the gap between the signal quality you have and the minimum you need. A 10G lane had enough margin to swallow a dirty connector without anyone noticing; a 25G lane doesn’t, so the same connector now causes errors. Margin is invisible on a healthy-looking link, which is the whole problem: it has to be measured rather than inferred.

MPO / MTP Connector
One connector housing many strands of fiber at once — typically 12, of which a 100GBASE-SR4 link only uses 8. MPO is the generic push-on interface; MTP is a trademarked, higher-performance version of it made by US Conec. It’s what makes parallel-fiber links practical, and it means one reseating mistake can affect several lanes at once.

Multimode and Single-Mode Fiber (MMF / SMF)
Multimode has a wide glass core and is used for short runs inside a building; single-mode has a narrow core and carries a signal much further. OM4, for example, is a multimode grade rated for 100G to about 100 meters — a rating on the glass, not on the current condition of your connectors.

NIC (Network Interface Card)
The network adapter inside a server. A 100G NIC only delivers 100G if everything behind it in the server — the PCIe slot included — can keep up.

NRZ (Non-Return-to-Zero)
The simplest way to send bits over light: on for a one, off for a zero, one bit at a time. Both 10G and 25G lanes use it, but the 10G lanes inside a 40G link run it with far more headroom than the 25G lanes inside a 100G link do.

OTDR (Optical Time-Domain Reflectometer)
An instrument that fires a pulse of light down a fiber and times what comes back, mapping every connector, splice, bend, and break by distance. It rules the cable plant in or out — necessary but not sufficient evidence that a link will actually carry 25G per lane.

PAM4 (Pulse Amplitude Modulation 4)
A way of encoding two bits per symbol instead of one, using four signal levels instead of two. It doubles the data rate over the same signaling rate as NRZ — which is how 25G electrical lanes become 50G and 100G optical links — but it also shrinks the margin between one signal level and the next. That shrinking margin is exactly why FEC becomes mandatory at these speeds instead of optional.

PCIe Gen3 x8
The server slot a network card sits in, described by generation and lane count. That combination tops out near 63 Gbps of real throughput — so a 100G NIC in one of these slots cannot actually reach 100G, no matter how healthy the network on the other end of the cable is.

Polarity, Type A / Type B
Polarity is making sure each transmitting strand lands on a receiving strand at the far end. Type A and Type B are standardized wiring methods that differ in how fiber positions are arranged inside the cable; neither is “correct” on its own, but the cassettes, trunks, and patch cords in a single channel all have to agree, or the link simply won’t come up.

Spine-Leaf
A network design where every access (leaf) switch connects to every core (spine) switch, and leaf switches don’t connect to each other. The leaf-to-spine uplinks are shared paths, so one degraded lane on one uplink can affect a lot of applications at once.

Strand
One individual fiber: a single thread of glass carrying light in one direction — except when using a BiDi (bidirectional) transceiver, which, despite what your instincts about light might tell you, sends signals in both directions over the very same strand at once. An SR4 link uses eight strands, and any single one of them can degrade without the link ever going down.

WDM / DWDM (Wavelength Division Multiplexing)
The general technique behind CWDM: stacking multiple signals on one fiber using different wavelengths of light. DWDM (Dense WDM) packs those wavelengths much closer together than CWDM does, which means more channels per fiber, pricier optics, and typically long-haul or carrier use rather than inside a data center.

Most of these terms only matter the moment something goes wrong — which is exactly when nobody wants to be looking them up for the first time. Keep this page handy.

Reliable Wi-Fi has become an essential part of the hotel guest experience. Guests expect fast, consistent connectivity that supports streaming, video calls, multiple devices, and the digital services they use throughout their stay.

For hotels, the right network can also support connected operations, guest engagement, and future technology investments.

In this white paper, you’ll learn:

• Why hotel Wi-Fi has become a critical part of the guest experience
• What guests expect from a reliable, high-performing hotel network
• How poor connectivity can affect guest satisfaction and online reviews
• Why capacity, coverage, and consistency matter as much as internet speed
• How reliable infrastructure supports hotel operations and connected technology
• What European hoteliers should consider when planning, upgrading, and managing Wi-Fi

Download the white paper to explore how a stronger connectivity foundation for European hotels can support better guest experiences today and the technology needs of tomorrow.

FULL ARTICLE HERE

Plain language connectivity questions answered in seconds. Increase Uptime

Wi-Fi monitoring with LLM integration is available today, and 7SIGNAL is among the first enterprise Wi-Fi platforms to offer it. Instead of logging into a dashboard to find out why a floor had high roaming-failure rates last Tuesday, ask your AI assistant. Instead of spending an hour building a weekly summary, prompt for one. Here is how it works, what you can do today, and where it is headed.

7SIGNAL makes its live performance data roaming, DHCP, signal quality, and EYERIS AI diagnostics available to AI assistants through the Model Context Protocol (MCP), the open standard Anthropic developed to connect LLMs to live external data. Any MCP-compatible assistant, including ChatGPT, Claude, and Copilot, can query that data to answer questions in plain language, generate reports, and drive agentic workflows grounded in what your network is doing right now, not in training data. It works with the AI assistant and the wireless infrastructure you already use.

“Our platforms allow our customers to extract, analyze, and act on Wi-Fi and other network performance data collected by our software agents and sensors. Now we’re making this data available to LLM platforms including ChatGPT, Claude, and Copilot, which means that anyone with access to a 7SIGNAL solution will be able to leverage the power of some of the world’s most sophisticated AI platforms to understand and analyze network performance.”

— Eric Camulli, VP of Customer Success, 7SIGNAL

Why can’t an AI assistant answer network questions on its own?

Because it wasn’t there. LLMs are trained on historical data with a cutoff; they can’t tell you which device is roaming poorly on your hospital floor today, or what your DHCP offer times looked like during yesterday’s shift change.

LLMs are extraordinary reasoning engines. They analyze complex data, identify patterns, generate structured reports, and explain technical findings in plain language. What they lack is the present tense of your network.

7SIGNAL has exactly that continuously at the device level across close to 1,000 metrics. But that data lives in a specialized dashboard, and most of the people who need to understand network performance operations managers, clinical directors, warehouse supervisors, IT leadership never log in to it. The insight stays trapped. MCP bridges that gap.

What is the Model Context Protocol (MCP)? MCP is an open standard, developed by Anthropic, that connects LLMs to live external data sources through a standardized interface. The 7SIGNAL MCP server exposes performance data, roaming metrics, DHCP telemetry, signal quality, and EYERIS AI diagnostics as tools any compatible LLM can query, so the assistant gets current, environment-specific answers instead of guessing from training data.

How does the 7SIGNAL MCP integration work?

7SIGNAL runs an MCP server that exposes its performance data as a collection of discoverable tools. The AI assistant acts as the client: when you ask a question, it determines which tools are relevant, calls them, receives the live data, and reasons over it in its response.

Nothing is guessed or hallucinated. The assistant retrieves actual metrics from your environment grounded in what 7SIGNAL Agents and 7SIGNAL Sensors measured, not in what a model assumes Wi-Fi behavior generally looks like. Because MCP is a vendor-neutral open standard, it is not tied to any one platform. As the ecosystem matures and more assistants add native MCP support, your network data becomes available across all of them without additional integration work.

What can you do with it today?

​Three capabilities are available the moment the connector is live no add-on, no roadmap wait. They map directly to the core ways teams use network data: asking questions, generating reports, and surfacing anomalies.

​1. Ask your network questions in plain language

No dashboard, no filter configuration, no export. You ask; the assistant retrieves the answer from 7SIGNAL and responds. For teams that previously spent 20–30 minutes navigating dashboards to answer a stakeholder question, that is an immediate and significant reduction in reporting time. Example prompts that work today:

“Which devices on the third floor had the most roaming failures last week?”

“How did DHCP offer times trend during yesterday’s morning shift?”

“Which access points have the highest rate of client disconnections this month?”

“What does EYERIS AI flag as the top three issues across the manufacturing floor right now?”

“Compare signal quality between Site A and Site B for Android devices.”​

2. Generate performance reports on demand

What previously required manual dashboard work pulling data, building tables, writing summaries, and formatting for a non-technical audience now takes a single prompt. Ask for a weekly site performance summary and get a formatted, readable report grounded in your actual metrics for that period, ready to send to IT leadership. Reports that used to take an hour to build take under a minute to generate.

3. Surface anomalies before users notice

EYERIS AI continuously analyzes close to 1,000 metrics for anomalous patterns across your device fleet. Through the MCP, those signals become early warnings an assistant can act on: DHCP offer times climbing on a subnet for 36 hours, a driver-version anomaly across a fleet of scanners, a handoff failure correlating with voice-picking errors on one route. These alerts only surface today after a user calls the help desk. Now they surface before the call.

Does it work with agentic workflows, or just chat?

Both. The 7SIGNAL MCP works inside agentic workflows too, not only questions typed into an assistant.​

The connector exposes the same live performance data to any MCP-compatible agent, which means an automated, multi-step workflow can query the network as one of its steps, with no person in the loop. ServiceNow is the clearest example: an AI agent running there can pull the affected devices, the time range, and the EYERIS AI diagnostic recommendation when an incident opens and attach them to the record so the diagnosis is in the ticket before any user has reported a problem. 7SIGNAL is not only a data source a person prompts; it is a tool surface an agent can act on. Today that means enrichment, triage, and ticket creation with full diagnostic context. The trajectory is autonomous remediation, where the agent assesses severity, opens the incident, and, within a defined scope and approval gates, pushes the fix.

What does this look like in daily operations?

Representative workflows available today with the 7SIGNAL MCP connected to Claude, ChatGPT, or Copilot.

Monday morning brief. A network engineer asks for a summary of Wi-Fi performance across all sites from the previous week. The assistant queries 7SIGNAL, retrieves site-level data, and produces a structured brief on the best- and worst-performing sites, the devices with the highest roaming-failure rates, and any EYERIS AI flags from the period. Under 60 seconds.

Stakeholder question. A hospital IT director asks why nurse satisfaction scores on the third floor dropped last quarter. The engineer queries 7SIGNAL to correlate roaming performance, VoIP call quality, and EHR access latency for that floor and period. The assistant returns a plain-language explanation with supporting metrics ready to share with clinical leadership, no dashboard navigation.​

Incident triage. An alert fires that scanner performance on the warehouse floor has degraded. The engineer asks the assistant to query 7SIGNAL for the affected devices, identify the common factors: access point, driver version, time of day, and cross-reference with EYERIS AI. A root-cause hypothesis comes back in minutes rather than hours, with specific remediation recommendations.

Automated ticket (ServiceNow). A workflow monitoring 7SIGNAL data detects a DHCP offer-time spike on a manufacturing subnet that has been trending upward for 48 hours. It automatically creates a ServiceNow incident with the affected subnet, the time range, the most-impacted devices, and the EYERIS AI diagnostic recommendation attached — before any user has reported a problem.

What changes for the network team?

The operational impact falls into three areas, each of which frees up time for work that requires engineering judgment. The throughline is simple: less time gathering data, more time acting on it.​

Less time on reporting. Performance reporting is one of the highest-time-cost activities that produces the least direct value; the time goes into formatting and presentation, not analysis. Assistant-generated reports grounded in live 7SIGNAL data remove that overhead entirely.

Less time on initial triage. The first 30–60 minutes of investigating a Wi-Fi complaint is typically spent gathering data: which devices, which access points, which time window, what the metrics show. The assistant does that in seconds, presenting a structured summary and a root-cause hypothesis instead of a blank dashboard.

Visibility without new tooling. Operations managers, clinical directors, and IT leadership can query network performance in plain language through the AI assistant they already use — without access to the dashboard or training on it. Network performance becomes part of the operational conversation, not a specialist report delivered on request.

What comes next: from LLM queries to agentic remediation

The current integration of natural-language queries, automated reporting, anomaly alerting, and ServiceNow ticket creation is the first stage. The next involves agentic workflows that not only identify and report issues but take action to address them.

Learn more about 7SIGNAL and ServiceNow

​In practice, that means workflows that detect an anomaly in EYERIS AI output, correlate it with historical patterns to assess severity, create a ServiceNow ticket with full diagnostic context, identify the recommended remediation, and, where the action is within a defined scope of autonomous operation, behind approval gates, push the configuration change to the relevant network controller. All without a network engineer touching the keyboard.

This is not science fiction. The data pipeline from 7SIGNAL’s sensors and agents through the MCP server to the assistant is already operational, and the ServiceNow integration is live. The remaining steps defining agentic action scopes, building remediation workflows, and establishing approval gates are engineering and governance work, not architectural work.

“Ultimately, the goal is to prevent any network disruption, but intermediate steps could, for example, be early warnings and integrating assurance data with support ticket systems like ServiceNow. We’re quickly moving towards much improved automated network assurance using evolving forms of AI.”

— Eric Camulli, VP of Customer Success, 7SIGNAL

SAN JOSE, Calif. – August 21, 2026 – Synaptics Incorporated (Nasdaq: SYNA) today announced that Ken Rizvi has resigned from his position as Chief Financial Officer, effective immediately, to pursue another opportunity. Mr. Rizvi will remain employed in an advisory role through September 30, 2026 to assist with the transition of his responsibilities.

Synaptics has elected not to conduct a search at this time for a successor CFO due to the announcement on June 25, 2026 of the pending merger between Synaptics and onsemi. Rahul Patel, President and CEO, will serve as the principal financial officer through the closing of the merger.

Mr. Kermit Nolan, Synaptics’ former Chief Accounting Officer, is returning as a consultant to advise the internal team through the transition. Over his 20-year tenure, he held several roles at the company, including acting CFO.

“Synaptics has a strong and experienced finance organization, and we are confident in the team’s ability to ensure continuity through the pending transaction,” said Rahul Patel, President and CEO, Synaptics. “Our strategic priorities and focus on driving success across Edge AI and Physical AI markets remain unchanged. I would like to thank Ken for his contributions to our success over the past few years and wish him the best in his new role.”

For a while, many believed working from home was temporary. Today, it’s simply how millions of people work, day in and day out.

The home has become an office, a classroom, an entertainment hub, and increasingly, a place where AI-powered tools are becoming part of the daily workflow. That evolution is changing what subscribers expect from their broadband experience.

In my latest byline for Broadband Breakfast, I explored new research from Plume showing how work-from-home habits continue to shape broadband usage and what that means for Internet Service Providers. The findings reinforce a simple reality: subscribers no longer judge their broadband service by speed alone. They judge it by whether it keeps them productive, connected, and uninterrupted throughout the day.

When working from home first became widespread, much of the industry’s attention focused on bandwidth.

Could networks handle the demand?

Would upload speeds be sufficient?

How much capacity would households need?

Those were important questions; today’s questions are different.

Can a video meeting remain stable while someone else is streaming?

Can AI applications respond instantly and successfully?

Can multiple users work, learn, and collaborate simultaneously without noticing the network at all?

The conversation has shifted from delivering connectivity to delivering confidence.

Our recent research with Parks Associates shows that subscriber expectations continue to evolve.

People rarely celebrate a faster speed test. They notice a frozen video call, poor audio quality, or an unresponsive app when they need it most.

That’s why the industry’s focus is expanding beyond throughput to Quality of Outcome (QoO). Success isn’t measured by how fast the network is in ideal conditions. It’s measured by whether subscribers can accomplish what they set out to do.

For remote workers, that outcome might be leading a customer presentation without interruption.

For someone using AI throughout the day, it might be receiving responses instantly enough to stay productive.

For families, it means all of those experiences happening simultaneously.

The next generation of work-from-home applications won’t simply consume bandwidth.

They’ll depend on low latency, consistent performance, intelligent traffic management, and continuous optimization.

As AI assistants become integrated into everyday workflows, subscribers will expect networks to adapt automatically to changing conditions.

Meeting those expectations requires more than additional capacity.

It requires intelligence.

Every connected home is different: The number of users changes. The devices change. Applications compete for network resources in different ways throughout the day.

Delivering a consistently great experience increasingly depends on understanding those changing conditions and adapting in real time.

That’s where orchestration becomes critical.

By continuously learning across more than 600 million connected devices, over 50 million homes, 450+ ISP partners, and 58 countries, Plume can recognize patterns that individual networks cannot. Those insights help providers optimize performance proactively rather than reacting after subscribers experience problems.

The biggest takeaway from this research isn’t that people are still working from home.

It’s that work from home has fundamentally changed what subscribers expect from broadband.

The providers that succeed won’t simply offer faster connections; they’ll deliver networks that are reliable, intelligent, and able to adapt to the realities of modern digital life.

For a deeper look at the research and what it means for broadband providers, I encourage you to read my full article in Broadband Breakfast.

Read the full byline: What ISPs Need to Know About Work-From-Home Habits.

Part One of this research paper covers the technical investigation: how the infection works, how Popanet operates and what traffic flows through subscriber homes without their knowledge or consent. Part 2 will examine how different types of malware are exploiting the presence of already-deployed proxies on these devices. All indicators of compromise are published on Plume Security Lab’s GitHub.

Subscribers buying a SuperBox believe they are paying for content. What they are also buying, without knowing it, is a permanent seat on someone else’s network.

Read Now

What happens when internet connectivity is interrupted at a hospital in the middle of running live diagnostics? What about a retailer processing thousands of transactions? Or a global conference streaming live to millions? In each scenario, business stops. Internet access becomes the single point of failure for the applications and services that keep the business running.

When the internet is the primary path to your applications, a connection failure isn’t just an IT issue. As the standard for connecting branch users and applications, direct internet access (DIA) requires resilient connectivity and robust security to ensure uninterrupted productivity and uptime.

AI-driven workloads, cloud services, and distributed teams are pushing more business-critical traffic directly onto the internet. As that traffic grows, the demand for performance and resilience accelerates.

Bringing networking and security together at the edge, Cisco 8000 Series Secure Routers are built to meet this demand. Independent validation from Miercom confirms that the 8000 series delivers, providing high-performance DIA with integrated security services in a single platform.

Validating DIA with integrated security

Miercom evaluated Cisco 8000 Series Secure Routers on two fronts: how much traffic they can handle and how well they stop threats while doing it. Miercom tested the routers across simulated small, medium, and large branch environments that mimicked real-world enterprise architectures, including cloud applications and SD-WAN controllers.

Across the portfolio, the platforms sustained throughput ranging from 1.0 Gbps to 9.5 Gbps while running next-generation firewall, intrusion prevention, URL filtering, application control, advanced malware protection, and network address translation (NAT) simultaneously. These results demonstrate that high-performance connectivity and comprehensive security can coexist as enterprises scale AI, cloud, and software-as-a-service (SaaS) applications.

Miercom also validated the platform’s integrated security capabilities. The Cisco 8235 Secure Router blocked 98.3% of evaluated attack strikes while sustaining approximately 2 Gbps of bidirectional throughput under load. Combined with identity-based microsegmentation, these capabilities help enterprises securely scale AI and cloud connectivity without sacrificing performance.

Figure 1. Cisco 8235 Secure Router results from Miercom’s independent testing

Make DIA your foundation

Increasing volumes of internet-bound application traffic are reshaping enterprise networks, making DIA a foundational service for modern branches. Cisco 8000 Series Secure Routers combine high-performance routing, integrated security, embedded visibility, and built-in post-quantum cryptography (PQC) in a single platform, simplifying operations and helping organizations securely scale.

• High-performance DIA: Enables secure access for cloud and bandwidth-intensive applications
• Converged networking and security: Simplifies operations with one integrated platform
• Greater cyber resilience: Protects users with advanced security and PQC-ready encryption
• Operational efficiency: Accelerates troubleshooting with embedded visibility
• End-to-end visibility and resiliency: Reduces hardware complexity with embedded ThousandEyes application-aware monitoring and automated cellular failover

 The next step

As internet traffic becomes more distributed, bandwidth-intensive, and security-sensitive, enterprises need more than fast connectivity. Edge operations demand high-performance routing, consistent security, and end-to-end visibility. Integrating these capabilities in a single platform makes this manageable. Miercom’s testing confirms that Cisco 8000 Series Secure Routers deliver everything the AI-driven enterprise needs, providing a trusted foundation for secure DIA while helping organizations simplify operations and prepare for future traffic demands.