On August 8, I took a trip to Yuhang District in Hangzhou. My destination was an OPC community called XX Bay—an offline hub built around the concept of a “One Person Company.” The event was themed around OPC cross-border compliance, covering things like payment integration.

The ticket was 29.9 yuan, the taxi there cost 64, and the subway back was 16. Altogether, it came to just under a hundred yuan. I went in with the expectation of “learning something real.” This post is partly a record of what I saw and heard, and partly some reflections of my own.

The Talk: Pretty Shallow, and Questions Hit a Wall

To be honest, there wasn’t much substance.

Most of what the speaker covered stayed at the surface level. The moment you tried to dig a bit deeper—“How exactly does this payment channel pass KYC?” “How do you isolate tax liabilities across different entities?” “Where are the risk-control trigger points for fund repatriation?”—he would shut it down with: “That’s a trade secret.”

At one point, another attendee asked a concrete question, and he directly quoted a price: paid consultation, 200 yuan for half an hour. Sitting there, I thought to myself: I already spent 29.9 on the ticket plus 64 yuan on a taxi to get here, and now if I want to ask one more question, I have to pay again?

There were no staff members, no materials, just one person reading from a PPT in front of a projector. The 29.9-yuan ticket didn’t even come with a bottle of water. Later, though, the OPC manager stationed there did find me a bottle—probably because I kept saying, “I’m so thirsty, is there any water around?”

A Sharing Session? More Like a Product Pitch

Later on, the speaker threw out a link: waffo.mintlify.app, and introduced a third-party product called Waffo Pancake. The pitch was that once you integrate with it, all your cross-border payment compliance problems are solved.

After I got back, I looked into it. Waffo is essentially a Merchant of Record service. It acts as the “legal seller” on your behalf, handling payments, tax, compliance, and payouts. Its cross-border settlement runs through LianLian Global. The main site is waffo.com, and the docs are hosted at docs.waffo.ai.

That explains why he kept saying “trade secret” whenever people pressed him. The real bottom line of the entire “knowledge-sharing” session may have been just one sentence: integrate with Waffo, and let it handle the rest. If he explained the underlying compliance logic too thoroughly, he would basically be undermining his own position: the audience would realize that the so-called “cross-border compliance expertise” was essentially “knowing this product exists and bringing it to you.”

So, for 29.9 yuan plus taxi fare, I bought myself an entry ticket to a product promotion event. That’s not necessarily an insult—packaging “use my product” as “let me teach you something,” then monetizing again through “paid consultation,” is itself a mature business model. It’s just that as someone who went there expecting to “learn,” the experience felt misaligned.

FDE Came Up On-Site: The Concept Was There, but the Implementation Details Were Thin

There was another detail at the event: several people were discussing the concept of “FDE,” saying it could be a transition path for traditional professions under pressure.

I looked it up afterward. FDE (Forward Deployed Engineer) is a hybrid technical role that has emerged in the era of AI foundation models and agents. Its core positioning is to serve as a bridge for “on-site implementation and end-to-end delivery,” specifically addressing the pain point where AI models are “easy to demo but hard to deploy.”

• Core responsibilities: Stay close to the client’s business environment, translate vague business needs into implementable technical solutions; handle the 0-to-1 development, private deployment, system integration (connecting with ERP/CRM, etc.), launch, and debugging of AI Agents and LLM-based systems; ensure that AI systems run stably in real business scenarios and quantify ROI, taking responsibility for final business outcomes.
• Required capabilities: Full-stack engineering skills (Python/Go, etc.) + AI engineering (building RAG systems, model fine-tuning, agent orchestration) + system integration + business understanding + client communication and cross-functional collaboration.
• Origin: The role was first systematized by the data giant Palantir, which deployed engineers directly to client sites to solve complex business problems. Later, as demand for real-world deployment of AI foundation models exploded, the model was widely adopted by OpenAI, Anthropic, major tech companies, and AI startups.

FDE differs from traditional on-site implementation consultants, who lean more toward process configuration, and from pure backend developers. Its essence lies in “understanding the business, writing code, getting systems online, and being able to take responsibility when things go wrong.”

To be fair, the people discussing FDE at the event weren’t clueless. Some of them did mention concepts like RAG, model fine-tuning, and Agent orchestration, and they knew these belonged to the FDE skill stack. But when you pushed further—“How exactly do you build it?” “How do you actually get it running inside an enterprise?”—there weren’t many concrete details. People had the conceptual understanding; what was missing was the implementation layer.

That was a reminder for me: concepts spread quickly, but the engineering capability required to turn those concepts into deliverable systems is much higher than it sounds. That’s also why FDE is a scarce role—it requires not just “knowing,” but “being able to do.”

It Wasn’t a Wasted Trip: I Got a Clearer Sense of the Offline Circle’s Actual Level

If measured by “how much cross-border compliance knowledge I learned,” this trip was a loss. But if measured by “getting a clearer view of the real level of this offline circle,” then the hundred yuan or so was worth it.

I also want to be honest about my own judgment here. I’ve occasionally said before that I’m “probably among the earlier movers in the AI circle”—but to be honest, there’s no objective ranking to stamp that claim as valid, and saying it always sounds a bit like bragging. Still, sitting in that salon and observing both the speaker and the audience, I could at least confirm one thing: the things I’ve actually gotten running, and the boundaries I think about day to day, are indeed a minority in that offline room. I’m not necessarily in some absolute “top 1%” ranking, but the statement “not at the mass-market level” roughly holds up.

That judgment isn’t something worth showing off. It’s more of a reminder: the overall level of this offline circle is even more beginner-stage than I had imagined.

A More Practical Byproduct: The Format Is Worth Referencing

Complaints aside, this salon gave me a fairly practical signal: the format itself is worth referencing.

One topic, one venue, a 29.9-yuan ticket, zero staff, zero material costs—and just like that, you gather people in. The ticket revenue from a single afternoon may be more direct than some digital products I spend days messing with. There’s nothing inherently wrong with the business model; they’ve validated it.

If I were to do something similar, the differentiator wouldn’t be “I’m better,” but rather that I have a working technical stack and real implementation experience, enough to keep going even when people ask deeper questions. Multi-provider proxies, AI agent orchestration, automated content pipelines—these are things I’m genuinely using, not just concepts I talk about.

So the next step is fairly clear:

1. First, consolidate—turn the things I’ve already gotten running into systematic sharing materials, rather than scattered verbal experience.
2. Then, consider speaking publicly—the format can be referenced, but the content has to be something I can stand behind. The barrier to entry isn’t the main point; what matters is whether the material can withstand follow-up questions.

I’ve always been clear about my own positioning: I’m not among the absolute strongest group, but when it comes to “having actually gotten things running,” I’m not empty-handed. The gap is not “whether I have something to talk about,” but simply “I haven’t started doing this yet.”

One-Sentence Wrap-Up

I didn’t retain much cross-border compliance knowledge from this salon, but it helped me see a few things clearly: the offline AI circle is more beginner-level than I expected; so-called “knowledge sharing” is sometimes just product pitching in disguise; and I do have things that are already working—I just haven’t started talking about them yet.

For a 29.9-yuan ticket plus nearly a hundred yuan including transportation, buying these confirmations was worth it for me.