Kenny Shultz

Before any slab cuts, we need to confirm the sanitary stub depth, verify slope to kitchen/bath layouts, and understand what that 1.5" water line is actually feeding. Small details, big consequences. Getting this right upfront avoids redesigns and mid-construction surprises. More Field Notes coming. #Construction #Engineering #MEPDesign #FieldNotes

We’re building tools that let Mac users interact with Revit models without ever opening Revit. Now imagine this: You’re cooking breakfast, talking to your agent through your headset: “Update the split system schedule to match today’s cut sheet.” “Create new light fixture types from VE fixtures.” “Update sequence of operations on M3 for the new DOAS controller.” “Tell the GC they should-” …well, maybe not that one. Just another day building what feels obvious now that we can talk ideas into existence. @PermitZIPelec

@awalehadam @MatthewBerman I want it to actually physically move my mouse and type on my keyboard though

@MatthewBerman the jump from 'llm in a box' to 'llm that actually moves the mouse' is the inflection point everyone was waiting for. spreadsheets are about to get a lot more interesting.

Should I even say it?

@ChShersh It makes sense. People who couldn’t build before suddenly can. That feels like a superpower, and it’s going to shift how they think about what’s possible.

I feel that people who are bad at coding are more bullish on AI. And this explains a lot.

I get your point about people chasing AI like they did NFTs trying to make quick money. That definitely exists. I just don’t think it’s the same. Buying a $2M ape isn’t comparable to spending money to build something. Even at the low end, people are learning, building skills, and creating things. That wasn’t really true with NFTs.

@PermitZIPhvac You’re bringing up totally unrelated things to my main point. Straw man logical fallacy.

Man it’s actually crazy how much people are spending on AI. Bc they feel like they are going to be left behind. But then they are not producing actually economically valuable outputs. It feels just like the FOMO of NFTs from 2021. Different form, same thing.

I don’t see the comparison to NFTs at all. This is people getting excited about finally building things they’ve been thinking about for years, and actually doing it. That hands-on experience is real and valuable. It builds skills. People spend far more on tuition every day, and for a lot of degrees, the practical return is honestly lower than what you get from actually making things.

@PermitZIPhvac consumer spending on AI tools is very different than business spending, so you missed the point. I am talking about all the people aping into vibe coding just like they aped into NFTs bc they thought they could make money from it.

@joshuamschultz @claudeai It’s probably the biggest overall productivity hack for my business and homeschooling and general life riff work. Streaming to a private database and wiring it up to mcp server that can do search is just like having a personal and business life repository.

@PermitZIPhvac @claudeai This is cool - didn’t think of this. Fireflies repo of so much weekly review help

My 10-year-old built three @claudeai artifacts this week as part of homeschool. A glossary. A quiz card deck. A learning simulator. Not because I told him to make an app… I told him to learn something enough to teach me, and that was how he chose to do it. Every week, above his regular subjects, I give him: - an AI project - part of a book I assign - a concept to teach me - a real course to work through (Stanford YT, Codecademy, etc.) The bar isn’t “finish the worksheet” or hours of meaningless homework. It’s learning marketable skills while building a real education.

@EricDoesDeals @joshuamschultz @claudeai We built the tool ourselves using deepgram speech to text api. Streams to a database. And then we wired up our own MCP server to get live interactions during the chat.

@PermitZIPhvac @joshuamschultz @claudeai What are you using to record, hardware and software?

@sweatystartup @grok @sweatystartup you’re assuming the $200/mo is the product. It’s not, it’s the on-ramp. The real model is consumption, and that’s far less subsidized. We’re already seeing it drive $200–$5k/employee in spend.

@sweatystartup @grok is consumption based token pricing the same delta?

People can’t handle the truth. It isn’t sustainable. Switching costs are virtually nothing. The bubble will pop when the VC subsidies stop.

@kylegawley Rage bait.

Anthropic has generated just $5bn in lifetime revenue. That's basically a rounding error for big tech. AI is a niche technology for nerds, nothing more.

@fhinkel It gets funnier every month. For every AI “gotcha” they point out, their own code likely has ten. And that gap keeps growing.

Senior engineers mocking AI agents for hallucinations is like blaming a toddler for not walking on day one. You don't dismiss potential because it isn't perfect yet.

@Helldemon @JenMcCready @aaronrubin If that were true, the token usage would back it up. That’s an easy thing to verify. You’re missing the forest for the trees here.

@PermitZIPhvac @JenMcCready @aaronrubin Who says they ignored the tools? They were down, do you understand the meaning of that? Maybe they already knew others had filed complaints, or maybe they decided to stick with their "older" faster methods.

I paused the company credit card used for Anthropic. Every employee that didn’t complain that Claude was down I fired.

@emadgnia @aaronrubin In 2024, we made a clean reset, and it was one of the best decisions we’ve made. We gave anyone who didn’t want to fully embrace AI the option to take three months’ salary and move on. Never looked back.

@aaronrubin This is the correct hiring filter for 2026. If your engineers aren't integrated enough to notice when their primary tool goes down — what are they actually building?

I wouldn’t want to be on a team where people ignore tools that clearly improve the work. If you’re serious about producing high-quality work, you use every advantage available. Some people are comfortable sticking with older methods, and that’s their choice. But it slows everyone else down. Teams that fully embrace AI move faster, make better decisions, and outperform those that don’t. And in practice, it’s hard to successfully mix those two approaches within the same organization.

@aaronrubin If this is real, it’s reckless. You disrupted work and then penalized employees for your own decision. That is not a culture anyone should accept.

@aaronrubin @LerWilson You already can, and we do. With cowork you can get all usage stats and can consume them to create your own dashboards.

@LerWilson Thank you sir, bringing this to Anthropic as an excuse to monitor all token usage on our corporate account.

@fidexcode Usually I can tell because I’m getting useful features and bug fixes into production immediately, even if they’re a little rough around the edges, instead of waiting two weeks for a Scrum process and a team of quarter-million-dollar engineers just to get it scheduled. lol

"How did you know I generated the code with AI?" The code:

@devXritesh @rezoundous but mcp tools (or tools in general) are basically agentic driven rag

@rezoundous This one also Tyler 👇 x.com/i/status/20349…

"SaaS is dead" "MCP is dead" "OpenClaw is dead" What's dead next, guys?

@rezoundous mcp definitely not dead. combining skills + agents + mcp is a powerhouse hack