I think OpenClaw is the reason we are getting so much AI slop these days. The comments here with "key insights" are coincidentally < 1 month old. It seems for some reason, OpenClaw/NanoClaw loves Show HN posts.
Yep. The 9B has excellent image recognition. I showed it a PCB photo and it correctly identified all components and the board type from part numbers and shape. OCR quality was solid.
Tool calling with opencode worked without issues, but general coding ability is still far from sonnet-tier. Asked it to add a feature to an existing react app, it couldn't produce an error-free build and fell into a delete-redo loop. Even when I fixed the errors, the UI looked really bad. A more explicit prompt probably would have helped. Opus one-shotted it, same prompt, the component looked exactly as expected.
But I'll be running this locally for note summarization, code review, and OCR. Very coherent for its size.
My friend works at one of the world's largest steel plants and recently he was discussing some ideas with me about solving the problem of people disappearing during inspections and being found dead after a few days. Apparently it is a huge problem there that he wanted to build a startup around the tech.
I am saying this because I think your target market may not be people stuck under rubble, but large scale industries, construction site workers, miners, firefighters, who can install the app beforehand. Cheers.
Thanks a lot for the insight! You definitely highlighted an aspect where this tool could be monetized to fund the development of an open source project like Igatha.
Sadly today, money and connections are needed for r&d and distribution. And having a market that buys a tool like this in certain circumstances can seriously contribute to finding/building publicly available solutions.
I’ve recently had to do a deep dive into construction tech + frontline (deskless) worker tech and there is a lot of potential in this space. Very underserved market. Lmk if you’d like to chat and I can provide some insight on what we’re seeing here in terms off greenfield opportunity
I would definitely like to chat about it. Right now, it's the main picture in mind towards getting some funding to push this project forward. So, I'd love to learn more about it.
I can imagine a firefighter, miner or steelworks worker relying on the smartphone in emergency. Like, seriously, a miner, half of a mile under the rock.... Carrying smartphone?
Heavy duty wearable tags and a dense network of wall-mounted relays pinging them and triangulating 24/7 can be the only solution. Sure you can use BLE beacons for that, and this is being solved with multiple competing products as we speak (look up "locating objects in warehouse"), but specifically for the human safety applications, they would have to be imposed by the regulators. Workforce is seen as disposable and nothing short of laws and regulations is going to force employers to improve safety.
The majority of the worlds miners work above ground - they still run the risk of stope (wall) collapse, but they're unlikely to find themselves under half a mile of rock.
In Australia, open-cut mining dominates coal production, accounting for over 80% of the total, while underground mining's share has decreased to around 15%. Open-cut mining is generally more efficient and cost-effective, allowing for higher recovery rates and larger-scale operations.
and
It has been estimated that more than two-thirds of the world’s yearly mineral production is extracted by surface mining. There are several types of surface mining, but the three most common are open-pit mining, strip mining, and quarrying. These differ from one another in the mine geometries created, the techniques used, and the minerals produced.
Ex miner here (grasberg, Indonesia) - yes, we all had smartphones, at all times, and yes, we rarely had signal but if you went up to the surface / anywhere with signal, you would want your phone, so we all carried them.
Are there any products that can act as a locator beacon, in perhaps a pager form factor? They listen for a signal from a powerful base station at frequencies low enough to deeply penetrate RF opaque material, and upon command, could key up and transmit at max power while using Time of Flight to triangulate (from a network of base stations)? 900mhz comes to mind, but perhaps there are alternate, more optimal frequency bands for this.
This is absurd. We have crossed the point of no return, llms will forever be in our lives in one form or another, just like internet, especially with the release of these open model weights. There is no bubble, only way forward is better, efficient llms, everywhere.
Won't a solenoid lock with microcontroller circuitry have a higher risk of failure due to water damage, power surges, and other factors(software bugs, hackers, cosmic rays!), in addition to wear and tear of a mechanical lock?
Well, it’s not such a big problem as it is in, say, health care.
It’s in the train co’s interest to have working toilets (not too many, because they displace seats but some) and have them reliable and easy to maintain (else costs go up, maybe even they need more per train). This is why they care about a denial of service attack (btw an electronic one can tell the conductor when the toilet is locked for a grossly atypical duration).
So to that degree the interest of the carrier and the passenger are aligned. This applies to most of the carriage decisions (robust seating, working doors and brakes, and so on).
But the if the trains pass first inspection, then it no longer is a development issue but a maintenance issue. [1]
It's not as if train companies can afford to run a statistically significant number of carriages from different suppliers to see which one gives them the actual best bang for buck 5 years down the road.
The short version: high speed trains failed and were eventually returned (2014) to the manufacturer for ~2/3rds of new price. (That's the exception part.)
There have not been high speed trains on this traject since. (That's the rule-conforming part.)
Source: been there, done that.
reply