tired kubernetes sre smiling - cartoon filter applied

dusty.ninja

Profile for bryan

Header for bryan
Avatar for bryan
Display name
Bryan :tng_badge:
Username
@bryan@dusty.ninja
Role
admin

About bryan

Fields

Cash.app
$jiqiren
Codeberg
https://codeberg.org/brybry
Github
https://github.com/jiqiren

Bio

ex-Apple, ex-Yahoo, ex-Digg, ex-NASA(JPL & Ames), “i like pop music”

Previously @Bryan

Kubernetes fanboi

Honolulu, Hawaii

Stats

Joined
Posts
2959
Followed by
478
Following
422

Recent public and unlisted posts

include boosts

I’ve read a number of mainstream reviews of people falling over themselves about how good #Muse is at being a personal assistant… so I thought I’d give it a spin.

To be honest, it is actually pretty bad for the task I gave it: find a USB-A to USB micro Type B connector. I think Muse’s inability is more because of all the anti-bot stuff going on.

#Amazon is really good at blocking bots, and so is #Cloudflare. Monoprice is behind Cloudflare, so Muse couldn’t even look at pricing. BestBuy and Walmart were viewable, but there were constant bot blockers that had Muse asking for help. I told Muse to solve the puzzles itself, and it was able to get past whatever Walmart and BestBuy do. After giving it the ok to purchase, it was a huge new pain in the ass. You either give it your personal account information or set up a new account. Then paying is some wallet connection with Stripe, which obviously won’t work everywhere. An alternative is a virtual screen running Chrome where you manually drive it over your phone. It kept triggering bot-blocking pages.  Since I was able to see the item in my cart for Walmart in their app on my iPhone, I was able to complete the purchase from a 3rd-party seller.

Overall, it’s kind of terrible at doing basic finding and buying. I shared this thought with Muse, and it agreed and stated it’s better at researching and finding.

So…. #AI #OpenClaw

1/2

Super weird. I just got a cold email from someone at IBM asking me to 'star' their project in the IBM namespace. It looks like an interesting project but it is totally not something I'd want to 'star'. 😑

I'll tell you what makes this interesting: this is software so AI Agents can control physical assets directly. Think HVAC systems... But at the same time manage physical assets that don't actually have a way for directly controlling them. Like you can't run tool from an AI and have trees trimmed or gutters cleaned. But sensors can be monitored and jobs can be dispatched to have reports filled out by meatbags (humans) so more work orders can be sent out.

So one of the arms of this MCP is creating, planning, scheduling, and approving work orders for humans to go out and make changes. That is, replace physical hardware or do a task.

This basically IBM amping up their existing software (Maximo) with AI.

Anyway, it is all Apache licensed which I guess is nice? or something?
#IBM #LLM #MCP
https://github.com/IBM/AssetOpsBench