we didn’t come this far, just to come this far.

Miami, FL
disposable software
GPT-6 and Intelligent UI, now rolling out in ChatGPT for everyone. Intelligent UI in ChatGPT delivers fast, interactive answers that make everyday questions more visual, complex topics easier to grasp, and interactive tools for your task available on the spot.
10
People will read again.
22
Is anyone on here an SI soloprenuer launching multiple new businesses at the same time? Follow up: Would you call yourself a soloconglomerate? Or is there another term we should be using? Thanks!
27
Agents are consuming more tokens than humans to create more agents to consume more tokens. So are these agents going to foot the compute bill? Cause last I checked us humans were the ones funding the buildout.
60
I beg your pardon…
Introducing Gemini 4 Argon – our new frontier model. It’s built for complex workflows across coding, enterprise knowledge work, and cybersecurity defense – rolling out today to a set of trusted testers through our Fairwind Program.
65
Big Beautiful Super Intelligence
Made with AI
2
164
Can't see how the iphone form factor survives in the agentic age.
47
Which would you rather be replaced by... 1) a bot 2) an instinct 3) a muse 4) a dot ?
87
What if the internet was never really ment for humans?
54
Assuming that the average corporate employee is going to find additional value for the enterprise when AI gives them an hour back from legacy workflows is naive. My bet is it goes to social media scrolling or other personal pursuits. And that’s why it has been so hard to measure AI ROI in the enterprise.
1
82
My AC has been leaking for a couple of weeks. Last Friday, Muse booked an HVAC technician to come service it this morning. They called to confirm details today. I missed the call. When I called back, they answered with an AI voice agent. Agents on either side. Compute on either side. The whole thing feels very bloated. I will report back when the service has concluded. It feels like this could have been done so much faster and less clunkily. But here we are.
152
AI agents like Muse use the human internet to get things done for us. That seems so inefficient. Who is building the agent internet? Is it just MCPs? Feels like we are both saying that compute is scarse and we need to accelerate compute infra but we are also so wasteful in how we are leveraging compute by slapping an agentic layer on top of the legacy digital world. Yes I can be bullish on SI and bearish on its current economics.
1
96
Yes, I can be bullish on SI AND bearish on short term SI economics.
81
“Technological innovation must drive global maxima, not local maxima.” @MatznerJon’s words got some teeth!
Here’s why Greg is wrong, from someone who bought, grew, and sold one of these: Nobody running of these is dumb, and the fact that they've survived is testament to the fact that they're smart. To me, baked into this strategy is a sense of paternalism. Look at these dumb boomers. I can come in with absolutely zero industry experience and a Claude Code subscription and change this industry!! The most pervasive dynamic within Main Street businesses is resource scarcity. Not enough time or money to do any of the million things that you plan to. Awesome, so you want to do some big Claude Code thing? Cool, you're going to need to find time to do that between figuring out how to repair the trucks, fix your broken Google My Business listing, and not take a salary that month because of the reinvestment you're making with the hope that it will pay off. Getting a CRM implemented correctly is hard enough!! You need managers who enforce the process, employees who change their habits, and someone accountable when things break. Now try rebuilding delivery around AI WHILE KEEPING THE BUSINESS ALIVE AND SERVICING DEBT. You come in full of piss and vinegar and on day four your operations manager quits because she liked doing things the old way, and suddenly you’re covering her job instead of building agents. Greg acknowledges key-person risk, but it deserves far more weight in his economics. The people, process, and change management required to make even basic technology productive can consume years. Innovation has to improve the performance of the whole business. Justin Roff-Marsh and I have discussed the sequence: identify the constraint, get more productive use from its existing capacity, then add capacity where necessary to increase throughput. Said another way ---If an accounting firm’s bottleneck is senior review, generating drafts ten times faster may just create a bigger queue. You have to show how AI frees the reviewer, change the workflow so that capacity gets used, and then upgrade whatever limits further growth. Technological innovation must drive global maxima, not local maxima. Even real savings elsewhere need to be converted into lower expenses (which is only weak replacement) or redeployed against the constraint. A list of automated tasks tells me very little about how much more profitable work the firm can complete. Now put a large, expensive acquisition loan underneath that transition. In the SBA buying range, debt payments compete with the money you need for implementation, training, replacement hires, and additional capacity. Saving people time doesn’t automatically remove their salaries. Your plan is to do what? Have Betty train you in the first three weeks to replace her, and then you're gonna fire her. Let me know how that goes for you with a big ass loan due every month. More than likely you're going to end up keeping Betty and adding a bunch of AI expenses, which are going to reduce all of your fancy pre-acquisition projections. You can spend months paying for the existing payroll, new software, and human review before realizing any savings.... If at all. And cutting a support role can make your scarce, expensive specialists less productive if they inherit its work. Ultimately a playbook like the below is achievable but requires an enormous amount of experience, judgment, skill, etc. And the idea that a corporate refugee is going to parachute in out of their W2 with a couple agents and transform a business model fundamentally feels like something that someone who's never tried to do it before in a real business would say. I'm happy to have this debate with anybody who'd like to have it. I've both owned these businesses and built agents in an enormous number of these categories referenced.
1
264
AJ retweeted
I have a feeling that a flood of AI related security events are about to hit the tape. Lets hope this post does not age well.
1
1
129