Gavin Campbell with Smart Yard Automation, Unraid 7.2 Enhancements, and the Rise of Local AI – HGG661

Posted on Sat November 1, 2025Sun August 2, 2026 by Jim Collison

Gavin Campbell from HomeTech.fm joins me to talk about what happens when smart-home technology moves beyond controlling lights and starts taking on larger jobs around the house.

We begin with Gavin’s experience using a Segway Navimow robotic mower and the ways automated mowing has changed his lawn-care routine. From there, we discuss the Echo Dot Max, his upgrade to Unraid 7.2 and several practical uses for local AI, including camera notifications, Home Assistant workflows, troubleshooting and software development.

The larger question running through the episode is where local intelligence fits as more products and services add AI. Some tasks benefit from the scale of cloud models, while others need the privacy, speed and control that local processing can provide. The most useful systems will probably combine both.

Audio

Audio Only

Gavin Campbell with Smart Yard Automation, Unraid 7.2 Enhancements, and the Rise of Local AI – HGG661

Video

HGG661 Chapters

[00:22] Welcome to Home Gadget Geeks
[00:49] Toronto Blue Jays and World Series Buzz
[02:44] Robot Lawn Mowers as a Practical Home Technology
[05:19] Navigation, Competition and the Future of Automated Lawn Care
[08:59] Gavin’s Upgrade to the Echo Dot Max
[16:17] Smart-Home Devices and Local Processing
[25:24] Upgrading the Home Server to Unraid 7.2
[31:36] Practical Uses for Local AI
[38:02] AI-Assisted Coding and Everyday Technology Work
[44:06] Is the AI Market Building a Bubble?
[51:28] Hardware, Data Centers and the Future of AI
[56:20] CES Plans, Personal Updates and Closing Thoughts

About Gavin Campbell

Gavin Campbell is a co-host of the HomeTech Podcast at HomeTech.fm, where he joins Seth Johnson and T.J. Huddleston for weekly conversations about home automation, consumer electronics, networking, connected products and the people building the smart home.

Gavin approaches home technology as both a longtime enthusiast and someone actively using these systems. His projects regularly include Home Assistant, home servers, networking, cameras, local AI, smart-yard equipment and the ongoing work required to keep a connected home useful.

Connect with Gavin

HomeTech Podcast at HomeTech.fm

What Is HGG661 About?

HGG661 is about the growing role of automation and local intelligence in everyday home technology.

Gavin explains how a Segway Navimow robotic mower has changed the way he maintains his lawn and what he has learned about navigation, traction, slopes and frequent automated mowing. We discuss how robotic lawn equipment is improving as manufacturers add better sensors and navigation while competition begins bringing more options into the market.

We also look at the Echo Dot Max and the way smart speakers are evolving beyond voice responses. Better audio is part of the story, but so are sensors, smart-home radios, network features and the gradual movement of some processing away from remote cloud services.

The second half of the episode moves into Gavin’s Unraid 7.2 upgrade and our use of AI for practical work. Local models can help analyze camera images, improve notifications, troubleshoot configurations and accelerate coding. Cloud AI remains useful for larger workloads, but local processing gives home-technology users another option when privacy, response time or direct control matters.

This Week on Home Gadget Geeks

Gavin and I begin with a little sports conversation before moving into the technology that had been doing real work around his home.

His Segway Navimow was one of the clearest examples. A robotic mower is easy to dismiss as a gadget until it takes over a repetitive task several times each week. Gavin describes what the mower does well, the adjustments he has made and the ways terrain still matters. Automated lawn care does not eliminate every problem, but it can change the amount of time a homeowner spends on routine mowing.

That conversation leads into the broader smart-yard market. Navigation is improving, more companies are entering the category, and technologies such as LiDAR are becoming more common. The important question is not whether every yard is ready for a robot. It is whether the technology is reaching the point where more homeowners can consider one without treating it as an expensive experiment.

Inside the house, Gavin had also upgraded to the Echo Dot Max. We discuss its improved audio, integrated sensors and place within a larger smart-home environment. Smart speakers increasingly act as combinations of voice interfaces, radios, sensors and network devices rather than simple countertop speakers.

Gavin then walks through his move to Unraid 7.2. The upgrade made the interface feel faster and more responsive and continued Unraid’s evolution as a flexible platform for storage, media applications, containers and home-lab workloads.

The rest of the episode focuses on AI. Rather than discussing AI only as a general trend, we look at what it can already do inside a real home. Gavin uses local models to help interpret camera images and create more descriptive notifications. We discuss AI-assisted coding, troubleshooting and configuration work, along with the tradeoffs between models running on local hardware and models accessed through cloud services.

Full Show Notes

Automating a Repetitive Lawn-Care Job

Gavin’s experience with the Segway Navimow begins with a simple value proposition: the mower can cut the lawn frequently without requiring him to schedule and complete the job manually each time.

Frequent mowing changes the process. Instead of waiting for the grass to grow and removing a large amount at once, a robotic mower can trim smaller amounts on a regular schedule. That can make the lawn look more consistent while reducing the amount of time Gavin spends actively mowing.

The mower still needs a suitable environment. Slopes, traction, obstacles, wet conditions and the physical layout of the property can affect its performance. Gavin has made adjustments to improve traction and help it work more reliably in the areas where his yard presents challenges.

The practical lesson is that robotic lawn care is automation, not magic. The mower can take over much of the repetitive work, but the homeowner still needs to understand the property, configure the system and pay attention to the conditions that cause trouble.

Better Navigation for Robotic Mowers

Navigation has become one of the most important areas of development for robotic lawn equipment.

Earlier products often depended on perimeter wires or relatively simple movement patterns. Newer systems can use satellite positioning, cameras, LiDAR and other sensors to understand where they are, recognize boundaries and avoid obstacles.

These technologies do not guarantee perfect operation in every yard. Trees, buildings, narrow passages, slopes and changing outdoor conditions can still complicate navigation. They do, however, make it possible for robotic mowers to handle more complex properties than earlier generations.

As more manufacturers enter the category, Gavin expects the available options to improve and prices to become more competitive. That does not make every mower inexpensive, but competition can move features that were once limited to premium products into a wider part of the market.

Living With the Segway Navimow

The most valuable part of Gavin’s experience is not a specification sheet. It is what happens after the mower becomes part of the weekly routine.

A robot mower has to coexist with the rest of the yard. That includes landscaping, slopes, changing weather, objects left on the lawn and areas where the mower may not have enough traction. Gavin discusses how he has worked through some of those details rather than treating the installation as a one-time setup.

That kind of iteration is common in home automation. The first configuration establishes that something can work. The later adjustments determine whether it becomes reliable enough to keep using.

The Navimow has been useful enough to reshape Gavin’s lawn-maintenance workflow. That is a stronger result than novelty. The system is doing recurring work that would otherwise require his time.

Echo Dot Max and the Expanding Smart Speaker

Gavin also shares his early experience with the Echo Dot Max.

The most immediate improvement is sound quality. A smart speaker still needs to function as a speaker, and better audio makes the device more useful for music, spoken content and everyday interaction.

The device also represents the continuing expansion of the smart-speaker category. These products can include microphones, environmental or presence-related sensors, smart-home radios and network features. Their value increasingly comes from how they participate in the rest of the home rather than from voice responses alone.

We discuss Matter support and the Echo’s relationship with Amazon’s Eero ecosystem. Depending on the rest of the network and the specific supported configuration, an Echo device can contribute to smart-home connectivity and network coverage as well as voice control.

The broader trend is toward combining more functions into devices that are already distributed throughout the house. That can simplify deployment, although it also makes the homeowner more dependent on the ecosystem that controls those functions.

Local Processing Inside Smart-Home Devices

Smart-home products have traditionally depended heavily on cloud processing. A command might leave the house, travel to a remote service, be interpreted and return before the device responds.

Cloud processing can provide access to more computational power and centralized services, but it also creates dependencies. Internet availability, vendor servers, account status, subscriptions and changing APIs can all affect whether a product continues working as expected.

Moving some processing locally can improve response time and allow basic functionality to continue when an external service is unavailable. It can also reduce how much information needs to leave the home.

Local processing does not automatically make a device private or reliable. The implementation, data handling, update policy and remaining cloud dependencies still matter. It does, however, give manufacturers another way to build smart-home products that are less dependent on a round trip to a remote server for every interaction.

Upgrading to Unraid 7.2

Gavin describes his upgrade to Unraid 7.2 as a noticeable improvement in responsiveness.

Unraid serves as a flexible foundation for his home-server environment. It can support storage, media services, containers and other home-lab applications without forcing every workload into a single fixed design.

Interface responsiveness matters because a home server is not merely installed and forgotten. Gavin uses the management interface to maintain services, inspect the system and make changes. A faster interface reduces friction during that ongoing work.

Upgrades still need to be approached carefully. A server may contain important data and support services used by other people or devices in the home. The benefits of new features and performance improvements need to be balanced against backups, compatibility and the possibility that an update could affect an existing workload.

For Gavin, the move to Unraid 7.2 produced enough improvement to make the system feel meaningfully better rather than simply changing the version number.

Local AI for Camera Notifications

One of Gavin’s most practical local-AI uses involves camera images and notifications.

A conventional camera alert may report that motion occurred. That can be useful, but it often leaves the homeowner to open the application and inspect the image or video.

A vision model can add context by examining an image and producing a description. Instead of receiving only a motion alert, the homeowner might receive information indicating whether the image appears to contain a person, vehicle, package or animal.

The model’s description should not be treated as infallible. Image-recognition systems can misidentify objects or miss important details. The notification is an interpretation that helps the homeowner decide what deserves attention, not a definitive security judgment.

Running the model locally can keep more of the camera workflow inside the home and reduce dependence on a vendor’s cloud-analysis service. It also requires enough local hardware and a model capable of performing the task within an acceptable amount of time.

Connecting Local AI With Home Assistant

Home Assistant provides a useful place to connect sensor data, camera events, automations and AI-generated interpretations.

The automation platform can detect that an event occurred, collect the relevant image or state information, send it to a model and use the response to create a better notification. This turns AI into one step in a deterministic workflow rather than giving it unrestricted control over the home.

That distinction matters. Home Assistant can continue owning the trigger, conditions and final action while the AI model supplies analysis or language. The model does not need authority over every device to make the notification more useful.

Local AI can also help during configuration. It can explain an automation, suggest changes, identify likely errors or help create the YAML and templates required for a project. The user still needs to test the result and confirm that the proposed behavior is safe.

AI-Assisted Coding

Gavin and I also discuss the way AI is changing software development and technical troubleshooting.

Coding assistants can generate boilerplate, explain unfamiliar code, suggest corrections and help translate an idea into a working first version. That can reduce the amount of time spent searching documentation or rebuilding common patterns.

The benefit is not limited to professional developers. Home-lab and Home Assistant users regularly write scripts, templates, configuration files and small applications. AI can help bridge the gap between knowing the outcome someone wants and knowing the exact syntax required to build it.

Generated code still needs review. A model can produce code that appears reasonable but uses an outdated library, assumes the wrong environment, introduces a security problem or simply fails to meet the actual requirement.

The most effective use is collaborative: the human defines the outcome and constraints, the model accelerates implementation, and the human tests the result.

GitHub Copilot and Local Models

GitHub Copilot represents the cloud-connected side of AI-assisted development, while locally hosted models offer another path.

Cloud tools generally provide access to larger models and managed infrastructure. They can be easier to start using because the user does not need to purchase, configure or maintain the hardware required to run the model.

Local models give the user more control over where data is processed and can work without sending every prompt or code sample to an external provider. They can also continue operating without a per-request cloud charge once the local infrastructure is in place.

The tradeoff is capability and maintenance. A model that runs comfortably on local hardware may be smaller or less capable than a leading cloud model. The user also becomes responsible for model selection, updates, storage, memory requirements and performance.

For many people, the best answer will be a hybrid workflow. Local models can handle repetitive, private or lower-risk tasks, while cloud models handle work that needs greater reasoning ability or a larger context window.

Privacy, Speed and Control

Local AI is often described as a privacy solution, but the real advantage is broader.

Processing data locally can reduce the amount of information sent to external services. It can also improve response time, remove per-request costs and give the user more control over model selection and system behavior.

Those advantages depend on the entire implementation. A locally running model may still receive information from cloud-connected cameras or interact with services that transmit data elsewhere. Local processing is one architectural choice, not a guarantee that the complete system is private.

Control may be the most important benefit for home-lab users. A local model can be selected, replaced, tested and integrated according to the homeowner’s requirements. That makes it possible to build workflows that are not entirely dependent on one provider’s pricing, policies or continued product support.

The Cost of AI Infrastructure

Our conversation expands from personal AI systems to the economics of the larger industry.

The most capable models require enormous investments in computing hardware, data centers, energy, cooling, networking and software. Companies are spending heavily because they expect AI services to become central to business and consumer technology.

That scale creates questions about sustainability. Demand may continue growing, but not every product, company or investment will succeed. The market could experience corrections as organizations discover which uses generate lasting value and which were built primarily around enthusiasm.

A correction would not necessarily mean AI disappears. Earlier technology cycles also experienced periods of excessive investment followed by consolidation. The infrastructure and practical uses that survive can continue developing after weaker business models fail.

Local AI is relevant to this conversation because not every task needs to reach the largest available model in a remote data center. Smaller models running closer to the user may handle some workloads more efficiently.

Local and Cloud AI Will Coexist

The conversation does not end with local AI defeating cloud AI or cloud AI making local models unnecessary.

Cloud models offer scale, managed infrastructure and access to capabilities that may be impractical to reproduce on personal hardware. Local models offer control, privacy options, predictable availability and the ability to integrate directly with systems inside the home.

Different tasks need different tools.

A brief camera description may be well suited to a local vision model. A complex research or reasoning task may benefit from a larger cloud model. A home-automation workflow may use deterministic local rules while asking an AI model to interpret only the unstructured part of the problem.

The strongest architecture is likely to route each task according to its requirements instead of forcing every workload through the same model or provider.

CES and What Comes Next

We close with Gavin’s plans for CES and the value of seeing new products and technologies in person.

CES provides an enormous view of the consumer-technology market, but not every announced product becomes successful or even reaches consumers. The useful work is separating practical developments from prototypes, marketing claims and ideas that are not yet ready.

For Gavin, the event is also an opportunity to connect with other creators and people working in home technology. Those conversations often provide more context than a product announcement alone.

I also share a personal update about Sammy’s library-science studies before we wrap up and invite listeners to continue the conversation in the community.

Key Takeaways

Gavin’s Segway Navimow has taken over a meaningful portion of his recurring lawn-maintenance work.
Robotic mowers still need configuration and adjustment for slopes, traction, obstacles and individual yard conditions.
Better sensors, positioning and LiDAR are expanding what automated lawn equipment can handle.
Competition should bring more robotic-lawn options and capabilities into reach for additional homeowners.
The Echo Dot Max combines improved audio with a growing collection of smart-home, sensor and network functions.
Local processing can reduce cloud dependence, but it does not automatically make a product private or reliable.
Gavin found Unraid 7.2 faster and more responsive in his home-server environment.
Local vision models can turn basic camera events into more descriptive Home Assistant notifications.
Home Assistant can retain deterministic control while using AI for interpretation, explanation and configuration assistance.
AI coding tools can accelerate home-lab projects, but generated code still requires testing and review.
Local models provide control and predictable availability; cloud models provide access to greater scale and capability.
Local and cloud AI are likely to coexist, with workloads routed according to privacy, cost, speed and complexity.
Heavy AI infrastructure investment may experience a correction without eliminating the practical uses that have already emerged.
The most useful AI projects solve recurring problems instead of adding AI simply because the technology is available.

Products, Platforms and Technologies Discussed

Segway Navimow

A robotic lawn-mower platform Gavin uses to automate routine mowing. The episode covers navigation, frequent mowing, slopes, traction and the adjustments required to make outdoor automation dependable.

Robotic Lawn-Mower Navigation

The category increasingly uses satellite positioning, cameras, LiDAR and other sensors to map properties, recognize boundaries and avoid obstacles. Performance still depends heavily on the individual yard.

Amazon Echo Dot Max

An Alexa-enabled smart speaker discussed for its improved audio and expanding role as a collection of smart-home, sensor and network capabilities.

Amazon Alexa

Amazon’s voice-assistant and smart-home ecosystem. The conversation considers how voice devices are evolving as more processing and connectivity functions are placed inside the home.

Eero

Amazon’s mesh-networking platform. Compatible Echo devices may participate in the Eero environment and extend coverage in supported configurations.

Matter

A smart-home interoperability standard designed to improve communication among devices and ecosystems. Matter support is part of the broader move toward multipurpose smart-home hubs and speakers.

Unraid 7.2

The home-server operating platform Gavin upgraded during the episode. He uses Unraid for storage, applications, containers, media and other home-lab workloads.

Home Assistant

The open-source home-automation platform discussed in connection with cameras, notifications, automations and local AI.

Local Language and Vision Models

AI models that run on hardware controlled by the user. The episode discusses their use for image interpretation, notifications, coding, troubleshooting and other home-lab tasks.

GitHub Copilot

An AI-assisted development service discussed as one example of using models to accelerate coding and technical work.

Cloud AI Services

Remotely hosted models provide access to large-scale computing and advanced capabilities without requiring the user to maintain equivalent hardware at home.

LiDAR

A sensing technology that uses light to measure distance and build an understanding of the surrounding environment. It is increasingly used for navigation and obstacle awareness in robotic equipment.

CES

The annual consumer-technology event in Las Vegas where manufacturers, developers, media and creators gather to demonstrate products and discuss emerging technology.

Frequently Asked Questions

What is HGG661 about?

HGG661 is a conversation with Gavin Campbell about the Segway Navimow robotic mower, the Echo Dot Max, Unraid 7.2 and practical uses for local AI in home automation, camera notifications, troubleshooting and coding.

Who is Gavin Campbell?

Gavin Campbell is a co-host of the HomeTech Podcast at HomeTech.fm. He regularly works with home automation, Home Assistant, networking, servers, smart-yard technology and local AI.

Which robotic lawn mower does Gavin use?

Gavin uses a Segway Navimow robotic mower.

Has the Segway Navimow replaced all lawn maintenance?

No. It has taken over much of the recurring mowing, but Gavin still needs to configure the system and account for traction, slopes, obstacles and changing yard conditions.

Why mow a lawn more frequently with a robot?

A robotic mower can remove a small amount of grass on a frequent schedule rather than waiting for the lawn to become long enough for a larger manual mowing job.

How does LiDAR help robotic lawn mowers?

LiDAR can help a mower measure its surroundings, improve navigation and identify obstacles. Its usefulness depends on how the manufacturer integrates it with the rest of the mower’s sensors and software.

What is the Echo Dot Max?

The Echo Dot Max is an Alexa-enabled smart speaker discussed for its improved audio and its role within Amazon’s broader smart-home and networking ecosystem.

Why did Gavin upgrade to Unraid 7.2?

Gavin upgraded to gain the latest Unraid improvements and found the management experience faster and more responsive after the upgrade.

What does Gavin use Unraid for?

He uses Unraid as part of his home-server and home-lab environment, supporting storage, media, containers and other applications.

How can local AI improve camera notifications?

A local vision model can examine an image associated with a camera event and create a description that provides more context than a basic motion alert.

Can AI camera descriptions be wrong?

Yes. Vision models can misidentify objects or overlook details. Their output should be treated as helpful context rather than a definitive security decision.

How does Home Assistant fit into the local-AI workflow?

Home Assistant can detect events, collect sensor or camera data, call an AI model and use the returned interpretation in a notification or automation.

Should an AI model directly control the smart home?

Not necessarily. A safer design is to let deterministic automations control triggers, conditions and actions while using AI only for analysis or language where uncertainty is expected.

Can AI help write Home Assistant automations?

Yes. AI tools can explain syntax, draft automations and help troubleshoot configurations. The resulting code should still be tested before it controls important devices.

What is the advantage of running AI locally?

Local processing can provide more control, reduce external data transfer, improve response time and avoid a per-request cloud dependency.

What are the disadvantages of local AI?

The user needs suitable hardware and becomes responsible for model selection, storage, updates, performance and integration. Smaller local models may also be less capable than leading cloud models.

Will local AI replace cloud AI?

Probably not. Local and cloud models have different strengths and are likely to coexist in hybrid systems.

Is there an AI investment bubble?

The episode discusses the possibility that some AI investments and companies may not be sustainable. A market correction would not necessarily eliminate practical AI tools or the infrastructure already being built.

What is the best reason to add AI to a home-technology project?

AI is most useful when it solves a specific recurring problem, such as interpreting an image, improving a notification, explaining a configuration or accelerating a technical task.

Join Jim Collison / @jcollison for show #661 of Home Gadget Geeks, brought to you by the Average Guy Network.

WANT TO SUBSCRIBE? http://theAverageGuy.tv/subscribe

Join us for the show live each Thursday at 8pmC/9E at http://theAverageGuy.tv/live

Find Us!

Join us in the Facebook group at https://www.facebook.com/groups/theaverageguy/

On Discord at https://theaverageguy.tv/discord

Get the Home Gadget Geeks subscribe links at http://homegadgetgeeks.com

http://theaverageguy.tv is powered by Maplegrove Partners web hosting. Get secure, reliable, high-speed hosting from people you know and trust. For more information, visit http://maplegrovepartners.com