Monday morning, I didn’t check the stock price first

At 9 a.m. Monday, I opened my laptop in a shared office in Nanshan, Shenzhen, ready to put together a one-page Claude product brief for a client. The client was pushing for it, but I saw the news first: Anthropic had gone public on Nasdaq at an IPO price of $83 and closed its first day at $97.17.

For a moment, I panicked. Not because I expected to make money from the stock, but because I suddenly realized that Claude—the tool I use every day—was no longer just a product from a research lab. As a freelance contractor, what I really care about is whether it will get more expensive, change its rules, and remain a safe place for my clients’ data.

What does this change mean for a one-person business?

Anthropic is the company behind Claude. The IPO hasn’t transformed ordinary users’ lives overnight, but quarterly results and business performance will now influence its product roadmap more visibly. At least for business customers, three questions will come up more often: Can I predict the cost? Will sensitive materials be protected? Is there a fallback if the service goes down?

I have a friend named Chenting who runs tax and accounting consultancy. In Hangzhou, she organizes records for more than 20 small businesses and used to check contracts and invoices by hand. Last month, after a client meeting near West Lake, she used Claude to extract dates, amounts, and penalty clauses from a contract, then reviewed every item herself. As she put it, AI saved her the “first read,” not the final responsibility; she still confirms every tax decision.

Another person I know is Liran, who is scaling a team. He runs customer training in Guangzhou and started using Claude last year to organize student questions and draft course outlines. He doesn’t upload student records directly. He first removes names, phone numbers, and company names. I also got this wrong: in the beginning, I uploaded whole client spreadsheets before realizing that anonymizing data means confirming whether every column is necessary—not merely deleting the header row.

If I wanted to recreate this setup, I wouldn’t start by “betting on Claude”

Here’s the math I think is more relevant to a small team: spend $20 on one month of Claude Pro, spend about one hour setting up Claude Code, and reserve two hours each week to test, refine prompts, and check results. That totals roughly 12 hours. The technical barrier isn’t high, but I need to understand file-upload limits, plan restrictions, and a basic verification workflow. The first step isn’t creating a paid account. It’s opening Claude’s official plans page, choosing the individual plan, and checking the payment amount and renewal terms before deciding whether to subscribe.

If I’m still using only the free version, I can postpone the purchase. I could turn one anonymized real task into a repeatable template and test it five times. There’s no urgency to try today; recording the most time-consuming steps may be more useful. Before uploading anything, I ask myself two questions: Is deleting names and phone numbers enough? Does the output need a second reviewer or a check against the original source? That matters more today than chasing IPO news.

How I’d choose based on the stage I’m at

If I’m a freelancer just getting started: I’d use the free version or a small monthly plan and test one repeatable task, such as turning meeting notes into a to-do list. I’d review it weekly to see whether it actually saves time, rather than rushing into a more expensive plan.

If I already have one or two stable clients: I’d build an anonymization process before deciding whether to pay. I’d test contract summaries, quote drafts, and customer replies separately, while keeping human review in the loop. Once an output becomes part of a client deliverable, I always compare it with the source.

If I’m scaling: I’d put Claude into a formal workflow instead of relying on personal experimentation. I’d track cost, response time, error types, and the share of output requiring human review, while preparing a backup tool. An IPO doesn’t mean Claude is necessarily better for me, but I’d rather work out the budget and fallback while the business is still small.