“`html

How Claude’s New Features Hook You Before You Notice

Anthropic’s latest Claude update isn’t just smarter AI. It’s a dependency machine. The new memory and Projects features retain context across every session, building a profile of you that grows more valuable the longer you stay. Anthropic is worth $61.5 billion according to Bloomberg. They’re not building a chatbot. They’re building infrastructure you can’t leave.

What Anthropic Actually Built

Claude’s Projects feature lets you store documents, instructions, and conversation history so Claude remembers your work style across sessions. You can build separate Projects for different clients, different workflows, different voices. The AI uses everything you store to give you more relevant answers every time you return.

Claude’s memory feature goes further. It learns your preferences across all conversations, not just inside a single Project. Over time, the Claude you’re talking to in month six is noticeably better at helping you than the Claude you met in month one. And that gap is almost entirely made of data you provided for free.

That sounds like a gift. It is, partly. But it’s also a trap. Every conversation you have inside a Project makes Claude more useful to you specifically. It also makes switching to a competitor more painful. You’d lose all that context. You’d have to start over. Most people won’t do that. That’s the whole plan.

The Numbers Behind the Play

According to Reuters, Anthropic surpassed $1 billion in annualized revenue in 2025. The Pro tier costs $20 per month. The Teams tier runs $30 per user per month. The business model only works if users stay and upgrade. Personalization is what drives both outcomes.

According to McKinsey’s 2025 State of AI report, more than 70% of companies that adopted AI tools planned to increase AI spending in the next 12 months. Fewer than a third had any formal plan for managing vendor dependency. That gap is exactly what Anthropic is selling into, and they know it.

I’ve tracked my own usage. I spend about 30 to 40 minutes per week maintaining and updating my Claude Projects: uploading templates, refining instructions, adjusting context. That’s time I’ve invested in Anthropic’s product. It’s time I’d lose if I left.

Most users see this as productivity. Smart operators see it as a subscription they can never cancel.

I’m not saying this is wrong. I use Claude Projects every day and it’s genuinely good. But good and dependency are not the same thing. You can get real value from a tool and still understand that the tool is designed to keep you paying. That’s the rich mindset. Poor mindset says “this is helpful.” Rich mindset asks “who profits when I find this helpful, and am I getting a fair share of that value?”

If you’re a freelancer or small business owner, the stakes are higher. You’re building workflows on someone else’s infrastructure. If Anthropic raises prices, changes terms, or kills a feature, your operation takes a hit. Smart operators hedge. They document their prompts, back up their Projects, and maintain a parallel workflow they can activate if needed.

If you’re formalizing a business around AI tools, setting up an LLC is one of the fastest ways to protect yourself and turn your software costs into deductible expenses. Inc Authority offers free LLC filing for new businesses, and it takes less time than you’d expect. Do it before you start signing client contracts or putting AI subscriptions on the business account.

What I Would Do Right Now

First, audit what you’re actually using. If you have a Claude Pro subscription but you only use it for occasional writing, you’re not getting enough value to justify the dependency risk. Either commit fully and build real workflows inside Projects, or stay on the free tier and keep your options open.

Second, document everything you put into Claude Projects. Your system prompts, your templates, your custom instructions. Store them in a Google Doc or a Notion page. This takes 20 minutes and means you can port your workflow to any competitor in a day if you need to.

Third, if you’re using AI for any work you do for clients, get your agreements in writing. A tool like signNow makes it fast to send, sign, and store electronic signature agreements with clients. As AI automates more of your output, clients will want clarity on what they’re actually paying for. A signed agreement protects you when that conversation happens.

Fourth, pay close attention to Anthropic’s enterprise moves. They’re building API integrations, agent features, and connections to business tools that will eventually lock in the enterprise contracts that drive serious revenue. If your company is evaluating AI vendors right now, understand that the vendor with the deepest integration is hardest to replace. Negotiate for data portability before you sign anything.

The Bottom Line

Claude is genuinely good. That’s the point. Anthropic isn’t tricking you with bad software. They’re hooking you with great software, and great software at scale is a business, not a favor. Every minute you spend teaching Claude your preferences is a minute you’re investing in their retention strategy. Use it, benefit from it, but keep your prompts documented, your agreements signed, and your business structured to survive any vendor change. That’s how you stay in control.

Frequently Asked Questions

What is Claude’s Projects feature and how does it work?

Claude’s Projects feature lets you store documents, instructions, and conversation history so the AI remembers your context across sessions. You can create separate Projects for different clients or workflows. The AI uses everything stored in the Project to give you more accurate and relevant answers every time you return.

Does Claude’s memory feature actually get smarter over time?

Yes, but only because of the context you provide. The more specific information you put into a Project, the more tailored Claude’s responses become. This is what creates switching costs: your stored context is valuable, and leaving means starting over without it.

Should small business owners worry about depending on Claude?

Not worried, but aware. Claude is a useful tool and using it well is a real competitive advantage. The risk is building critical workflows on a single vendor without any backup plan. Document your prompts, keep your data portable, and structure your business so you can adapt if pricing or terms change.

Is Anthropic a stable long-term vendor?

Anthropic is solidly funded, with a $61.5 billion valuation as of 2025 and major backing from Google and Amazon according to public filings. But the AI market is still early and competition is fierce. Don’t build a business that can only survive if one vendor stays dominant.

How can small businesses reduce AI tool costs?

Track usage versus value every month. If you’re paying $20 to $30 per seat, each seat should save at least 10x that in time or revenue generated. File your business as an LLC so subscriptions are deductible, and ask for annual pricing discounts since most AI vendors offer 15 to 20% off for annual commitments.

“`