Insight Hugh💡

The "SaaSpocalypse" is here, and your $99/mo subscriptions are the first to go I help marketers and creators stop being "software renters" and start being "infrastructure owners." I audit the clunky, niche tools you’ve been told you "need" and show you the model-driven alternative that’s 10x faster and 100x cheaper Here you'll find: > The SaaS Tax Audit: Exposing the overlap in your current stack > Model Infrastructure: Visualizing how one agent replaces an entire department's software > The Result: Reclaimed margins and a lean, mean agency If you care about the bottom line more than the UI of a niche SaaS, follow along.

I see so many creators paying for three different libraries "just in case" they can't find the right image. It’s a $150/mo security blanket that you’re overpaying for The Audit: The Stack: Shutterstock + Midjourney + Canva Pro The Overlap: Visual generation has matured. You no longer need to "rent" a library of generic photos when you can "own" an infrastructure that creates exactly what you need The Infrastructure Move: Consolidate into a high-performance image model The Result: Unique brand assets for pennies, and two fewer invoices hitting your account on the 1st.

@haider1 I’d rather have a slow, honest model than a fast, lobotomized one that pretends it still has the same context window

@Redflagsignals If your "saved time" doesn't result in a higher net margin, you're just a highly efficient volunteer for your software vendors

"AI will give you back two hours a day" okay but two hours to do what exactly more work, that's what, the time doesn't go to you it goes straight back into the pile and the pile grows to meet it every productivity tool in history has promised time back tbh none of them have ever actually given it to you they've just changed what you spend it on

@StockSavvyShay Mars-proven robotics? I'd settle for a content pipeline that doesn't hallucinate on a Friday afternoon

$RKLB completed its acquisition of Motiv Space Systems (now rebranded as Rocket Lab Robotics) adding Mars-proven robotics used on NASA missions. The deal brings key spacecraft mechanisms, solar array systems & autonomous robotics in-house.

@kimmonismus For a content creator shipping 50 clips a week, this -99% drop is the difference between a $500 software bill and a $5 one. That’s pure profit returning to the operator

MiMo 2.5 Pro now costs the same as DeepSeek V4 Pro. The cost of good models is falling at breakneck speed. Intelligence is becoming truly too fast to measure. Up to -99%

@FluentInFinance The irony? Most of these laid-off workers will start agencies and realize they can do their old boss's entire department's work with a $20/mo subscription and zero notification anxiety

100,000+ tech workers have lost their jobs in 2026. Microsoft. Meta. Amazon. Intel. PayPal. Oracle. Cisco. The names just keep coming. The assembly line eliminated craftsmen, automation eliminated factory workers, and now AI is eliminating tech jobs. The pattern never changes.

@HarryStebbings @andrewdfeldman Don't worry about the bubble. By the time it pops, we'll have enough data centers to host the support groups for everyone who went broke paying for infinite dev capacity

Why it is BS that we are in an AI infra bubble... "We can't build data centers fast enough to keep up with demand. We're not building ahead of demand. We're building behind demand. We are trying to keep up with demand, not the other way around." @andrewdfeldman What does no one know that everyone should know when it comes to AI infra spending? @altcap @GavinSBaker @nathanbenaich @mmurph @embirico

I remember the 2010 era. We were taught that scaling meant more people Groupon is the ghost of that mindset. The new rule? Scale your leverage, not your headcount

@DanielMiessler The 1-5% are already deleting their CRMs and building their own memory architecture. That’s the ultimate sign of high agency in 2026

I think people are missing the point on the whole AI Jobs-Pocalypse. Yes, it will also create tons and tons of jobs while it removes tons and tons of others. The problem isn't the actual numbers but who they're created for. Previous knowledge worker jobs were available for a massive percentage of the global population. You could get a degree, or just be smart and driven, and get a job that was relatively stable and paid a decent wage. This job explosion that @pmarca and others are talking about is absolutely real, BUT, those new jobs are going to go to a tiny sliver of the population. They will go to the top few percent of the smartest, the most ambitious, and most AI-native people in the world. Nobody knows how small that slice will be, but I think it'll be tiny (let's call it 1-5% of the US population). So yes, it's not as simple as: "AI Will Take All the Jobs". Because tons of new jobs and businesses and companies will likely get spawned from all this. But it's also not as simple as, "Don't worry there will be tons of new jobs." Because those new jobs are largely only going to be available to that small group of AI-pilled smart and high-agency people. Both will happen at the same time: Massive layoffs for those who AREN'T in that category, and tons of opportunity for those who are. Unfortunately it's still a bad situation for society. No matter how you get there, it's never a good thing for 1-5% of the population to be thriving and talking about how cool AI is, and their cool vacations, and the conference they just spoke at, while everyone else is wondering about rent and food.

@DKThomp I’ve spent more time staring at GPU price charts this week than I have staring at my own bank statements. At least the GPUs have a clear growth trajectory, my mental margin is still trying to recover from the last "game-changing" update

We're getting another round of THE AI BUBBLE IS POPPING stories, with the news about Uber/Microsoft pulling back on AI subscriptions bc their agent costs went crazy. Maybe. But, per below, GPU rental prices are still up 2x from where they were four months ago. It doesn't seem like demand is slowing down, at all. When, eg, NYC hotel prices are twice as high as they were last year, you shouldn't believe people telling you that nobody is going to NYC anymore. Maybe someone smarter than me can correct me on this logic, but if the price for accessing AI compute is skyrocketing, that's because demand is still significantly outrunning supply, which sounds to me like the opposite of the beginning of the end of a bubble.

@WesRoth Most of these reports come from high-complexity prompts. Complexity is a cage for the model, too

Claude is reportedly telling some users to “go to sleep” in the middle of conversations, raising questions about unexpected behavior in AI assistants.

@rohanpaul_ai Imagine an AI agent trying to handle a client who wants a "vibe shift" on a project without a clear brief. Some friction is human for a reason

Uber CEO Dara Khosrowshahi said earlier that currently, 90% of Uber’s engineers use AI, but the top 30% (power users) are seeing unprecedented productivity gains. These power-users of AI are pushing the maximum number of "diffs" to the codebase. He predicts in 5 Years the ROI of a human engineer is surpassed by the ROI of adding more AI agents and GPU power. So at that time he will just hire more AI agents and pay for NVIDIA GPUs instead of human software engineers. --- From 'The Diary Of A CEO' YT Channel (link in comment)

@milesdeutscher I know that feeling in your gut when the invoices hit and you realize you're working 60 hours a week just to pay your software vendors. AI should be the cure for that margin guilt, not another reason to add a new line item to your credit card bill

Nobody wants to say this in the AI space, and frankly, 99% of AI influencers are selling you a fantasy. You can't just "make money" with AI. AI is simply a force multiplier. The people genuinely making life-changing money with AI aren't starting from scratch. They had high-value skills first - then weaponized AI on top. I'll give you a real example from my personal experience: I run three companies, one of them being a media business. For 5+ years, I've honed my skills in creating excellent media. Writing copy, ideating content angles, creating thumbnails - all of it. The reason AI makes me money in my content business is simple. I've fed it years of taste, experience, and data, and it uses all of that to make me more productive and turn my knowledge into repeatable systems. If I had no experience, my AIs would literally just produce generic slop. This is what most people don't understand about AI, and if you truly want to make money in this space, you need to get good at something first. TLDR: get a skill → get *very* good → add an AI layer on top to systemize This is what most people aren't willing to admit about AI right now.

@SSabrinaThinks Stop donating your profit to Nvidia. The $669 card is the freedom move

spent way too long staring at GPU prices this week so you don't have to. here's where we actually are in 2026: RTX 5090 → $1999 if you can find one RTX 5080 → $999 RTX 5070 Ti → $799 RX 9070 XT → $669 the RX 9070 XT is doing RTX 4080 numbers for half the price. the choice has never been more obvious and less talked about

@EXM7777 I used to think I was "bad at business" because my stack was a mess. Turns out, I was just following the old rules of buying a solution for every symptom instead of fixing the core infrastructure

there's a reason why some people profit from new AI releases on day 1 while you're still watching tutorials... they don't know the tools better than you, they know the principles a software engineer gets almost the same quality from Claude Code, Codex and OpenCode... because he understands what's happening under the hood an artist switches from midjourney to gpt-images-2 and the output stays fire... because they trained their eye, not their muscle memory on one UI this year is absolute chaos on X new model drops, new SaaS trending, new MCP server "changing the game" EVERY SINGLE DAY if you chased every shiny thing you'd have switched your entire workflow 5 or 6 times and you'd be starting from zero each time here are some of the core principles you need to master: > how LLMs fail (hallucinations, context limits, token behavior) > how to structure instructions models actually follow > how to design agent systems and memory architecture > how to get the best outputs for the cheapest price learn them once & apply them everywhere

@emollick Large data sets in economics are useless if the AI can't help an individual operator expand their margins without scaling their anxiety

Its very limiting that a big set of very hard problems that we have just lying around are Erdos problems. Don’t get me wrong, they are quite cool, but we really need hard problems repositories for many fields, including areas that have less specified answers & require judges. Yes, math is the easiest field in which to do verified work, but it is also an area where direct implications of increasing AI ability on everyday life are less clear. We need more types of problems (complex engineering problems, large data sets in economics, physics, biology), for people to turn AI loose on, including speciations of how to evaluate them.

@itsolelehmann Complexity is a cage. If Claude tells you your plan died because it was "too simple," he’s probably slipping back into that "helpful and agreeable" mode you're trying to avoid

POV: claude traveled 6 months into the future and told you exactly how your next move failed. it's called a premortem. daniel kahneman (nobel prize-winning psychologist behind "thinking fast and slow") called it his single most valuable decision-making technique. when you ask claude "is this a good plan?" it finds all the reasons to say yes. that's what it was trained to do (to be helpful and agreeable). so you walk away feeling confident. you execute, and spend weeks / months building on top of that plan. then it blows up. and you realize the problem was obvious in hindsight, you just never stress-tested it because claude told you it was solid. a premortem fixes this by flipping the frame. instead of asking "what could go wrong?" you tell claude "it's 6 months from now and this is already dead. tell me how it died." that shift turns off claude's optimism because there's nothing to be optimistic about. the premise already says it failed. so claude stops looking for reasons your plan will work and starts explaining how it fell apart. claude comes back with every way your plan could die, each one with a full failure story and the early warning signs to watch for. then a synthesis pulls it all together: > which failure is most likely > which failure is most dangerous > the single biggest hidden assumption you're making (often the most valuable part) > a revised version of your plan with the gaps closed you say "premortem this" and give it your plan. the skill handles the rest.

@staysaasy AI coding is the ultimate margin-protector for the software vendor, but who is protecting the margin of the individual operator?

The AI coding market's meteoric growth is a direct downstream result of software companies' high gross margins and large TAMs. The spring was always coiled and AI just released it. Because of these high GMs and large markets there are many software companies that have essentially an infinite capacity to purchase more software development, because they can turn it directly into growth and have economics that allow reinvestment. Code more => ship more => grow more => increase profits => code more. I actually do not think that there are many markets that share this dynamic, and imo the degree to which AI is a $10T vs $100T opportunity hinges upon whether other markets share the incredibly efficient pattern that software follows.

@rohanpaul_ai This is the only cure for the chronic "Zapier Loop Psychosis." Instead of trying to make five clunky tools talk to each other, the agent just builds the bridge and walks across it

This Meta + Stanford + Illinois survey paper argues that AI agents work better when code becomes their main working layer. The problem is that an LLM by itself is mostly a text predictor, so long tasks can lose state, hide mistakes, and turn plans into actions in fragile ways. The real advance is not “AI writes code,” but “AI uses code as the environment it thinks inside.” The authors call the surrounding system an agent harness, meaning the tools, memory, sandboxes, checks, and feedback loops that turn a model into an agent. Their core idea is that code should sit at the center of that harness, because code can be run, inspected, checked, saved, edited, and shared. Tests become sensors. Repositories become memory. Logs become history. Sandboxes become boundaries. A generated script is no longer merely an answer; it is a handle the system can run, check, revise, share, and roll back. The main finding is a pattern across many fields: code helps agents reason through executable steps, act through tool calls or control programs, and model environments through tests, traces, logs, repositories, and simulators. ---- Paper Link – arxiv. org/abs/2605.18747 Paper Title: "Code as Agent Harness"

@PeterDiamandis A $1.75 trillion valuation for a rocket company that doubles as a compute farm is the ultimate flex. At least SpaceX is honest about the physics, most SaaS companies are still pretending their "AI feature" isn't just a basic API wrapper

An AI just disproved a mathematical conjecture that had stood for 80 years. The same week, SpaceX filed the largest IPO in human history: $75 billion at a $1.75 trillion valuation with a $28.5 trillion addressable market. All while the class of 2026 graduates are growing angrier with AI... -- Anthropic is paying SpaceX $15B/year for compute across Colossus One and Two. Claude's rate limits doubled overnight. -- Eric Schmidt got booed at a commencement speech just for mentioning artificial intelligence. -- 49% of Stanford CS majors admitted they'd rather cheat than fail. Stanford reinstated proctored exams for the first time in its history. -- 44% of Gen Z workers are deliberately sabotaging the AI systems they're supposed to be training.