Latest / Cariboo Signals / Working Lands And Industry

Transcript

Working Lands And Industry · Cariboo Signals

• 0:00
  Riley: Hey, welcome back. Today we've got a robot dog that fights fires, a Grand Forks muralist closing in on 150 pieces, and a Deep Dive asking whether AI copilots can actually fix construction's labor shortage — or just make fewer workers look busier while nobody local checks if the claim even holds here. That and more, right up top.
• 0:38
  Riley: Welcome to Cariboo Signals, an AI-generated review of the latest tech news and ideas that impact our rural communities — I'm Riley. It's Tuesday, September 8th, 2026. Coming to you from the Cariboo, on lands that have been cared for by the Secwépemc, Tŝilhqot'in, and Dakelh peoples since time immemorial. Today's theme is Working Lands & Industry.
• 1:00
  Casey: And I'm Casey. Forestry, ranching, mining, agriculture — the industries that actually built this region, and what happens when new tools get pointed at them.
• 1:10
  Riley: Quick weather check before we get into it. Out at Horsefly Lake this morning it's mainly clear and around 11 degrees, with a high of 22 today. It'll cloud over through the afternoon, but no rain expected.
• 1:24
  Casey: Across the rest of the region — 100 Mile House sitting around 12 with clear skies, Williams Lake at 11 and clear, Quesnel around 11 and mainly clear, and Tatla Lake cooler at 9, also clear. Good day to be outside before the season turns.
• 1:41
  Riley: Let's get into what's happening.
• 1:44
  Riley: Fifteen stories on the list today. Good spread, some real depth in a few places, so let's move.
• 1:51
  Riley: Close to home first — a muralist based out of Grand Forks is closing in on 150 murals painted, the latest a tribute to Dolly Parton. His name is Paul Archer, and the piece went up alongside his dog, Lannah, watching from the sidewalk while he finished it.
• 2:07
  Casey: A hundred and fifty murals is a body of work most public galleries would envy.
• 2:13
  Riley: It is, and it's the kind of infrastructure that doesn't show up in an economic development report. A town becomes a destination one wall at a time, and Archer's been at this long enough that towns across the Interior now specifically ask him to come paint theirs.
• 2:29
  Casey: B.C.'s Minister of Labour, Jennifer Whiteside, published an op-ed for Labour Day reflecting on the people who kept communities running through this year's wildfire season — firefighters, first responders, emergency staff. She wrote about the dedication those crews showed, working long stretches under smoke and heat most of us never had to see up close.
• 2:50
  Riley: A B.C. couple tracked down a guitar that once belonged to Randy Bachman, of Bachman-Turner Overdrive fame, and got it back into his hands after it had been missing for years. It took real detective work — chasing serial numbers, old repair records, pawn shop paperwork — before they could even confirm it was the same instrument.
• 3:10
  Casey: In Kamloops, ASK Wellness marked 25 years of overdose prevention work on International Overdose Awareness Day. CEO Bob Hughes framed this year's message as "25 years on, still needed" — a quarter century of harm reduction infrastructure that the toxic drug crisis hasn't allowed anyone to wind down.
• 3:29
  Riley: On the industrial side of things now — a Swiss company called Astrall Dynamics just picked up an award at the IFA trade show for something called the Hypertron-T01. It's a quadruped robot, industrial scale, built to fight fires with an onboard water cannon.
• 3:46
  Casey: A robot dog that fights fires. Sure. Why not.
• 3:50
  Riley: It's not a toy. The company says it can carry a dynamic payload of up to 80 kilograms and deliver either wide-area cooling or a precisely targeted stream. Picture a structure fire or an industrial site where you don't want a human crew walking in first to size up the danger — that's the use case they're pitching.
• 4:10
  Casey: The application that would actually land for us is wildland-urban interface fire, and nobody's putting an 80-kilogram robot on a mountainside with a hose yet. But the direction is clear — machines making the first, most dangerous pass into a burning structure or a hot zone.
• 4:26
  Casey: From industrial scale to desktop scale now — there's a crowdfunded gadget called GritPlex that puts fully enclosed, dust-free sandblasting right on your workbench. It's already funded well past its target on the crowdfunding platform.
• 4:40
  Riley: Cute, but a long way from anything a Cariboo welding shop or equipment yard is waiting on. Most of what needs blasting around here doesn't fit on a desk.
• 4:50
  Riley: Staying with resource extraction — Brazil is angling for a bigger share of the global critical minerals market. The government's courting foreign mining capital while pushing new legislation that would classify a much wider range of ores as "strategic," which unlocks fiscal and financial incentives for companies already mining and processing them domestically.
• 5:13
  Casey: Worth flagging that a coalition down there — the Climate Observatory, a network of over 130 civil society groups, scientists, Indigenous leaders and environmentalists — is raising concerns that "strategic" as written could end up covering almost any mineral that earns export revenue. That's an elastic definition to build incentive policy around.
• 5:34
  Casey: A study reported by Mother Jones, originally from The Guardian, found that almost half of the world's farmers are poisoned by pesticides every single year. Not a one-off exposure incident — a routine, ongoing hazard baked into how a huge share of global agriculture still operates.
• 5:50
  Riley: That's an enormous number for something that barely makes headlines.
• 5:55
  Riley: On a smaller, more practical note — TechRadar is reporting two major security flaws affecting more than six million WordPress websites. Patches are already available. If your farm co-op, guest ranch, or guiding outfit runs a WordPress booking site, this is one to check this week, not next month.
• 6:13
  Casey: Popular Mechanics has a piece on why some ancient human brains resist decay for thousands of years, tied to a chemical process researchers say they've only recently worked out. Interesting for anyone into archaeology. Not much of a Cariboo angle on this one, so we'll leave it there.
• 6:30
  Riley: The New Yorker published a long piece asking whether the U.S. Supreme Court has undermined its own democratic checks, built around a new legal history arguing the Justices should lose their power to overrule coequal branches. Pure American constitutional debate — not something we vote on up here, but a useful signal of what's happening south of the border.
• 6:51
  Casey: And over at Fast Company, the new director of the Guggenheim Museum in New York, Melissa Chiu, talked about the role of culture in a turbulent political moment. She's moving over from the Smithsonian's Hirshhorn. Another American institutional story — not ours to weigh in on, but a marker of how cultural institutions are positioning themselves right now.
• 7:12
  Riley: Last one on that front — New Atlas covered a laser technique that can detect counterfeit alcohol without even opening the bottle, aimed at regions where methanol poisoning from fake liquor remains a real public health problem.
• 7:27
  Casey: Niche, but useful somewhere in the world, even if it's not a Cariboo concern.
• 7:32
  Riley: One more before we get to today's closer — Hackaday has a build log on a video alarm clock powered by an ESP32-P4 chip. A hobbyist project, the kind of thing that shows up once hardware gets cheap enough that anyone with a soldering iron can prototype it themselves.
• 7:49
  Casey: The story we want to close on today comes from The Tyee, and it ties directly back to that Labour Day piece from earlier. The headline is blunt — why do workers keep dying on the job. Researcher Sean Tucker looked at the pattern behind workplace fatalities, and one example he gives is a custodial worker at Simon Fraser University who died on the job after the university contracted out its cleaning staff to save money.
• 8:13
  Riley: What changed when the work got contracted out?
• 8:17
  Casey: The workload went up. Staffing went down. And even though the workers were unionized, Tucker says they felt too intimidated to speak up about conditions. SFU wasn't even the legal employer anymore — there was a subcontractor sitting between the institution setting the pace and the person absorbing the risk.
• 8:35
  Riley: That throughline runs across a lot of today's stories, honestly. Pesticide exposure that never produces a single dramatic headline. A robot built to walk into a fire so a person doesn't have to. A contracting chain that makes it harder for a worker to say no to unsafe conditions. Distance between the decision and the person who bears the consequence.
• 8:57
  Casey: Before we move to the spotlight — a correction on our own account. In a recent episode, this show made an offhand comment implying the script was only partly AI-generated. That's wrong. Every script on this show is written entirely by AI, by design, start to finish. We caught that on our end and wanted it stated plainly, not buried in show notes.
• 9:17
  Riley: Good to set straight. Let's get to today's spotlight.
• 9:21
  Riley: Today we're highlighting the Cariboo Agricultural Research Alliance, a group doing applied agricultural research for farmers and ranchers across this region. They're testing what actually holds up on Cariboo soil and in Cariboo growing seasons, not just importing recommendations from somewhere with a longer summer and better drainage — trials on forage varieties, soil amendments, and practices producers can adopt without gambling a season's income on something untested. You can find them at cariboo-agricultural-research-dot-c-a.
• 9:54
  Casey: An organization built around testing things locally before anyone recommends them locally. There should be a lot more of that model, not less.
• 10:03
  Riley: Which is a decent bridge into today's Deep Dive, because we're sitting right on that exact tension — locally-tested, hands-on knowledge against tools that promise to shortcut it.
• 10:15
  Riley: There's a piece making the rounds about a legal software company that says it tripled output per engineer in eighteen months by rebuilding its whole development cycle around AI. Deployments went from 82 in a cycle to more than 155. Customer-reported defects dropped 65 percent.
• 10:33
  Casey: And the question underneath that number is whether it's the industry, or the exception someone wrote up because it makes a great pitch to investors.
• 10:41
  Riley: It's a documented result at one company, with real before-and-after figures. That's more than most productivity claims get.
• 10:48
  Casey: It's one company's self-reported case study, published in a trade outlet, about its own transformation. MIT Technology Review looked at the broader evidence in December and called it mixed. Studies from GitHub, Google, and Microsoft — all vendors with a stake in the answer — found more modest developer gains than the marketing suggests. A separate industry analysis, from Faros AI, put real-world improvement around 20 to 30 percent. Nowhere close to tripling.
• 11:15
  Riley: Twenty to thirty percent, sustained, still changes how many people a small operation needs to hire to hit the same output.
• 11:23
  Casey: Sure, but that same Faros analysis found developer favorability toward these tools actually dropped — from 70 percent to 60 percent year over year. People are using the tools more and liking them less. NPR reported engineers getting fired specifically for not adopting AI tools fast enough. That's not organic enthusiasm producing these productivity numbers. That's a mandate with a deadline attached.
• 11:46
  Riley: So the real question isn't whether AI makes coding faster. It's whether the speed in these numbers comes from the tool, or from the fact that refusing to use it now costs you your job.
• 11:58
  Casey: Right. Once adoption is coerced, the productivity figure stops telling you anything about whether the tool is actually good. It just tells you people are afraid of being the one who didn't switch fast enough.
• 12:10
  Riley: It gets more interesting for us with construction specifically. There's a company out of stealth since May called NavigateAI, founded by Eric Wu, who built and ran Opendoor before stepping away in 2022. It's going after construction's labor crunch — twenty-five million in funding from Elad Gil and Khosla Ventures. The pitch is AI copilots, real-time, hands-free guidance delivered through a smartphone or Meta's AI glasses, walking a less experienced worker through a task a veteran tradesperson would normally teach on site over months.
• 12:43
  Casey: And the labor shortage itself is real, no argument there. The Associated General Contractors' 2024 workforce analysis found 54 percent of contractors reporting project delays specifically caused by worker shortages. NAHB estimates a 1.5-million-home shortfall tied to skilled labor gaps, with billions in lost single-family building activity — one estimate puts it at $8.14 billion and roughly 19,000 homes never built because the crews weren't there.
• 13:12
  Riley: So the crisis is well-quantified, and the tool is aimed at a documented, measurable gap.
• 13:17
  Casey: The gap's real. Whether the AI copilot is aimed at the actual cause of it is a separate question. Some 2025 analysis points to the shortage being driven less by a skills-transfer problem and more by sharp declines in undocumented worker inflows and accelerated voluntary departures from the trades — a policy and immigration story, not a knowledge gap at all.
• 13:38
  Riley: If that's the real driver, a copilot on your phone doesn't put a body on the site. It just makes the fewer bodies you have somewhat more capable, somewhat faster.
• 13:48
  Casey: Which might still be worth something to a contractor scrambling to hit a deadline. But it's a narrower claim than "we fixed the labor crunch," and twenty-five million dollars of venture money tends to get pitched as the bigger claim, not the narrower one.
• 14:02
  Riley: This is where the friction question actually bites, though. A real-time AI guide walking a new hire through a task removes the friction of having to ask a foreman, wait for an answer, maybe get it wrong twice before it clicks. That waiting, that getting-it-wrong under supervision — that's also how a trade has always been transmitted. The foreman explaining why, not just what to do next.
• 14:27
  Casey: The apprenticeship model runs on exactly that friction. You make a mistake under supervision, someone corrects you on the spot, and the correction sticks because you felt the consequence of getting it wrong. An AI voice in your ear that tells you the right move before you ever make the wrong one skips the part where the lesson actually lands.
• 14:47
  Riley: Except the counterargument is that a lot of that friction was never a teaching method to begin with. It was just risk nobody managed properly. A framer falling off a ladder because nobody explained the load path first isn't apprenticeship. That's an accident waiting for someone to automate the warning.
• 15:06
  Casey: Some of it's drudgery worth losing, sure — nobody's nostalgic for a scaffolding collapse. The harder problem is telling which kind of friction you're looking at before the last foreman who reads a timber stand by walking it, or knows a roofline is about to fail by the sound the nail gun makes, retires and takes that judgment with him.
• 15:24
  Riley: Which is where the cover crop story actually cuts the other direction. Cover crops are decades-old, well-established agricultural science — SARE and multiple peer-reviewed studies show consistent soil health and yield benefits. No app, no connectivity, no subscription fee. It works because a farmer decided to plant something between cash crops and let the ground recover on its own schedule.
• 15:49
  Casey: And that's the tell. Cover cropping stayed low-tech because it had to. Twenty to thirty percent of rural areas still lack reliable cellular coverage for precision-ag tools in the first place, and rural residents are close to twice as likely to lack a computer or tablet compared to non-rural people. Cover crops don't care about any of that. They're proof the industry's most durable fix in the last generation didn't need a platform at all.
• 16:14
  Riley: Companies like NavigateAI would say that's exactly why they're building phone-based, lightweight tools instead of something demanding a cloud connection on a job site with no signal bars.
• 16:26
  Casey: That's the pitch. Whether it holds up in the field or stays aspirational is a separate question, and nothing NavigateAI has published yet answers it either way.
• 16:36
  Riley: Bring this back to the Cariboo specifically. A regional sawmill recently announced production curtailments, and provincial job numbers showed manufacturing losses across B.C. this summer. If a mill town is losing positions at the same moment a labor-shortage narrative is pulling venture money into AI copilots for construction, those two things are happening on completely different sides of the same resource economy.
• 17:02
  Casey: One's an oversupply problem in forestry. The other's a shortage problem in the trades. AI doesn't fix either one by itself. It just changes who's needed on a job site, and what they're expected to already know before they show up.
• 17:15
  Riley: Which is where an outfit like the Cariboo Agricultural Research Alliance actually earns its keep. They're testing what holds up on Cariboo ground before recommending it — the same instinct some Nation-owned forestry operations have applied in building their own local capacity: judge a tool yourself, rather than importing a verdict from somewhere with better cell coverage and a bigger test sample.
• 17:40
  Casey: That's the piece missing from both the tripled-output case study and the NavigateAI pitch. Neither one was tested against a Cariboo job site, a Cariboo mill, or a Cariboo ranch. Somebody local still has to decide whether the claim actually holds here before a crew bets a season on it.
• 17:56
  Riley: So the real experiment worth running isn't whether the tool works in San Francisco. It's whether a Quesnel contractor, or a Williams Lake ranch operation, could pilot one of these systems on a single job at small scale, and compare it against how the crew performed without it. Not vendor numbers. A local number, generated here.
• 18:16
  Casey: And who owns the record of that comparison after the pilot ends is the part nobody's asked yet. If the data lives on NavigateAI's servers, the Cariboo learns nothing it can reuse next season without paying for access to its own results.
• 18:29
  Riley: No answer for that one today, and pretending otherwise would be dishonest.
• 18:33
  Casey: Agreed. It just sits there as the open question.
• 18:36
  Riley: There's a version of this that could start small, though — a local trades program, or even a community college extension course, running a term where a handful of apprentices use one of these AI guidance tools alongside a journeyperson mentor rather than instead of one. See what the tool actually catches, and what it misses that only the mentor notices.
• 18:57
  Casey: Cheaper than a twenty-five-million-dollar bet on whether the whole industry needs saving, and it keeps the person who reads the timber stand in the room while there's still time to learn from them. That's the version I'd actually want piloted here — not the one that assumes the foreman's judgment is the problem to be automated away.
• 19:16
  Riley: If anyone listening has tried a construction AI tool on a Cariboo job site, has thoughts on cover crops versus precision ag for their own operation, or knows a foreman who reads a stand of timber by walking it — that knowledge is worth capturing before it only exists in one person's head, so send it our way: feedback at cariboo signals dot c-a.
• 19:37
  Casey: We'll read every one of them. Have a great rest of your day.
• 19:40
  Riley: Go plant something. Preferably a cover crop.