Three years ago, Claude cost twenty dollars a month and felt like a magic trick. You could ask it anything, get a thoughtful answer, and run experiments all day without thinking about it.
That era is ending. Not loudly, not all at once, but unmistakably.
In March, Anthropic officially confirmed what Pro plan users had been complaining about for weeks: weekday peak-hour sessions burn through your usage allowance faster than they used to. Weekly caps — which didn't exist before 2025 — are now standard. And the "promotional capacity bumps" Anthropic keeps announcing? Those are short windows of generosity inside a tighter long-term contract.
Anthropic also raised API rate limits this spring, by as much as 1500% on the lowest builder tier. That sounds like a gift. It isn't.
Here's what's actually happening. Heavy users — especially anyone running Claude Code at scale — used to be able to do real work inside a $200/month Max subscription. As subscription caps tightened, those users have been pushed onto metered API billing instead. The API charges per token. The rate limit increase didn't lower the price; it just raised the ceiling on how fast you can spend.
We've seen this movie before
Twice in the last five years, actually.
Remember when Netflix cost $7.99 a month and had everything? That was streaming's cheap-tuition phase. It trained a whole generation to expect on-demand entertainment as a near-free utility. Then the catalog fragmented across Disney+, Max, Peacock, Paramount+, Apple TV+, and the average household bundle climbed past eighty dollars a month. The economics never made sense at the subsidized price — it was a customer acquisition cost the industry could only afford until interest rates rose.
Then there was the SPAC era. From 2020 to 2021, zero-percent interest rates flooded markets with cheap capital. Every experimental bet got funded. The 2023 reset killed the buffet. The people who built real intuition during the cheap years compounded an advantage that the late-arrival cohort never caught up on.
AI is in the middle of the same reset right now.
What this actually means for your business
If you've spent the last three years treating AI as someone else's problem — a toy, a novelty, a thing your nephew is into — the toy phase is over.
The cost of catching up just went up in three specific ways:
Per-experiment cost. When experimentation was effectively free, you could try ten ideas a day to figure out what worked. With weekly caps, tighter sessions, and overage that hits at API rates, every experiment carries more weight. You get fewer shots.
The intuition gap. Three years of cheap inference let early adopters compound learning — what to build, what to skip, when an AI tool actually saves time versus adds friction. That knowledge can't be bought retroactively. It can only be learned slowly, or borrowed from someone who already did the learning.
The compounding curve. The people building useful AI systems today are building on top of three years of working understanding. They're not starting at zero. You would be.
None of this means AI is inaccessible. It means the on-ramp got steeper, and the cheap, exploratory route most people would have taken is closing.
Here's the part that matters
You don't have to pay the tuition yourself.
I've spent the last three years building AI systems in production for real businesses. I've built complete go-to-market systems — website, email sequences, lead generation, all of it — in as little as five weeks. I have products shipping in market right now: MCPSkills, WayPoints, Switchboard, Cardvark. I build agents that handle real work, not theoretical demos.
What I do for small businesses comes down to three things:
Save you time. The tasks eating your week — email drafting, content production, customer follow-up, data entry, research — most of them can be automated or handed off to an agent that does them better than the rushed version you'd squeeze out at 11pm.
Grow your business. AI is genuinely useful for the parts of growth that small businesses can't afford to staff: SEO content engines, lead qualification systems, customer service that runs while you sleep, internal tools that let one person do the work of three.
Skip the learning curve. The decisions I've already made — what works, what's a waste of money, what scales, what breaks — those are the decisions you'd otherwise spend the next year making yourself, paying retail tuition the whole way.
Three lanes at Rise Above:
- Live classes for owners and teams who want to compress the learning curve.
- Products I've already built that you can use today.
- Custom builds — websites, content engines, agents, automations — tuned to your specific business.