We are upgrading the global markets of any sort during this decade. I'd not go linear and claim that "first came crypto and then AI". Though both cluster-of-sectors have been in the making for more than half a century—yes, that's correct and i like m-dashes—only in the last decade have people come to understand that they can manage their own money in the programmatic level easily.

Today, it's happening to AI, too. People who, yesterday, hated knowledge management tools—either proprietary or open source, e.g. Notion, Obsidian—are today building local context engines for themselves. Today, people come to learn they can steer their daily personal context through global markets.

The seasoned retail investor, the average VC, or the witty serial player all know that in times like this—consider the last two weeks of December 2025 where this tangible vibeshift started—know that you can routinely, and in full mediocrity, can ship "products and startups and companies" that can be easily shipped off by the likes of Claude, Cursor, OpenAI, NVIDIA as a mere plugin or addon and "make profits".

Soon, we are entering in that period across the following verticals: -

• stablecoins

• prediction markets

• LLM vibecoding adjacent B2B tooling

• personal knowledge and context management rails.

Whatever you are building will possibly, unless you are not Palantir, Anduril, Flexport, Tesla or anything similar, that is if you are not making chips, either, in fact, to be more clear, if you are not "manufacturing" or "assembling" but merely shipping non-hyperscale oriented ordinary tools that jazz with the AI age—cliché, right?—will be routed, and aggregated, just like how people finally cared enough to build DeFi liquidity aggregators, into main provider's desktop app, mobile app, terminal CLI and the like.

Your startup is already a standalone plugin/ skill in the nerd's tooling, and "your" startup will to be internalized, which is why it's kind of better to offer premium tools in plugin, MCP, data stream (x402 + ERC-8004 pipeline, too, looks promising). In the end, you'll be left with a sort of obsolete-over-a-week product website being served through enterprise Notion, lame fundamentals that any opinionated late-faux-geek can show off with in the leftover circles of 2015 Twitter and the current Bluesky.

Act slowly but fast. Burn ideas on a daily basis and context compress with whatever LLM you have access to. I have access to Claude Code Plus, ChatGPT Plus, and Cursor Max for a few months and I promised myself I'll build a product that will make the coming Summer funkier.

Now, this piece is not about me, though.

What's the argument here?

I keep seeing people encouraging founders go full "consumer", which is indeed a risky business unless you have a long runway, good funding—even those compute deals will still be expensive as a good jazz gig—and even the B2B tooling engineers do not understand where this is all headed. There might be business development, solution architecture and scaling + optimization focus teams (you can usually find these people under the so-called alignment depts for one reason or another, i guess it's easier to label that way) but few of these teams are well-incentivized in the face of the job they are undertaking to offer proper NDA'd forecasting models per these markets, and no, no two-comma AI-researcher will join your smol team in the foreseeable future.

So, you need to make sure that your product faces both the consumer end, preferably via a generous freemium front-end product, and your real deal should be operated as plugins, addons, packages, APIs.

Yes, you need to focus on the solar, the energy, the datacenter, the storage, the inference, and you can well cook a steak at home.

We are upgrading the global markets of any sort during this decade. I'd not go linear and claim that "first came crypto and then AI". Though both cluster-of-sectors have been in the making for more than half a century—yes, that's correct and i like m-dashes—only in the last decade have people come to understand that they can manage their own money in the programmatic level easily.

Today, it's happening to AI, too. People who, yesterday, hated knowledge management tools—either proprietary or open source, e.g. Notion, Obsidian—are today building local context engines for themselves. Today, people come to learn they can steer their daily personal context through global markets.

The seasoned retail investor, the average VC, or the witty serial player all know that in times like this—consider the last two weeks of December 2025 where this tangible vibeshift started—know that you can routinely, and in full mediocrity, can ship "products and startups and companies" that can be easily shipped off by the likes of Claude, Cursor, OpenAI, NVIDIA as a mere plugin or addon and "make profits".

Soon, we are entering in that period across the following verticals: -

• stablecoins

• prediction markets

• LLM vibecoding adjacent B2B tooling

• personal knowledge and context management rails.

Whatever you are building will possibly, unless you are not Palantir, Anduril, Flexport, Tesla or anything similar, that is if you are not making chips, either, in fact, to be more clear, if you are not "manufacturing" or "assembling" but merely shipping non-hyperscale oriented ordinary tools that jazz with the AI age—cliché, right?—will be routed, and aggregated, just like how people finally cared enough to build DeFi liquidity aggregators, into main provider's desktop app, mobile app, terminal CLI and the like.

Your startup is already a standalone plugin/ skill in the nerd's tooling, and "your" startup will to be internalized, which is why it's kind of better to offer premium tools in plugin, MCP, data stream (x402 + ERC-8004 pipeline, too, looks promising). In the end, you'll be left with a sort of obsolete-over-a-week product website being served through enterprise Notion, lame fundamentals that any opinionated late-faux-geek can show off with in the leftover circles of 2015 Twitter and the current Bluesky.

Act slowly but fast. Burn ideas on a daily basis and context compress with whatever LLM you have access to. I have access to Claude Code Plus, ChatGPT Plus, and Cursor Max for a few months and I promised myself I'll build a product that will make the coming Summer funkier.

Now, this piece is not about me, though.

What's the argument here?

I keep seeing people encouraging founders go full "consumer", which is indeed a risky business unless you have a long runway, good funding—even those compute deals will still be expensive as a good jazz gig—and even the B2B tooling engineers do not understand where this is all headed. There might be business development, solution architecture and scaling + optimization focus teams (you can usually find these people under the so-called alignment depts for one reason or another, i guess it's easier to label that way) but few of these teams are well-incentivized in the face of the job they are undertaking to offer proper NDA'd forecasting models per these markets, and no, no two-comma AI-researcher will join your smol team in the foreseeable future.

So, you need to make sure that your product faces both the consumer end, preferably via a generous freemium front-end product, and your real deal should be operated as plugins, addons, packages, APIs.

Yes, you need to focus on the solar, the energy, the datacenter, the storage, the inference, and you can well cook a steak at home.