AI Anxiety & Go-to-Market

Several participants converged on the same observation: AI has crossed a perceptual boundary. It no longer behaves like a faster version of something familiar.

The specific example: language models used to stream output at a pace you could read along with. You could follow the reasoning, catch mistakes, and feel like you were collaborating. Current models produce complete, sophisticated output in seconds. There's no following along. You either trust the output or you audit it after the fact.

This creates a dependency dynamic that the group found uncomfortable. The cargo cult analogy came up naturally: we're receiving gifts from a system we don't fully understand, and we're organizing our businesses around them.

The practical response: you don't need to understand how the models work internally. You need to understand what they're good at, what they're bad at, and how to structure your workflow so that their mistakes get caught before they cause damage. That's a design problem, not a technical one.

The Abstraction Ladder

Programming has always moved up layers of abstraction: machine code → assembly → C → higher-level languages. Each transition made the previous layer less visible while preserving the need to understand what you were asking the computer to do.

AI is the next layer. You don't need to think about variables, functions, or syntax in the same way. But you still need to know what you're building, why, and how to communicate that to the model. The skill shifts from writing code to articulating intent precisely.

What This Means for Founders

Two implications:

First: The barrier to building software has collapsed. Anyone with patience and a clear idea can produce a functional product. A 7-year-old with a 3D printer and AI-generated marketing collateral can plausibly start an Etsy business.

Second: Because everyone can build, building is no longer the differentiator. Domain expertise, customer relationships, and the ability to design something that actually solves the right problem are what separate products that get traction from products that just exist.

"Maybe not human coders, but human designers."

Energy as the Real Constraint

The fintech founder's primary concern wasn't model capability — it was energy. Training and running frontier models requires enormous amounts of power. China has invested more aggressively in energy infrastructure for AI. If the US falls behind on the energy side, it constrains everything built on top of it.

The Open-Source Paradox

China, which internally uses AI for surveillance and social control, is externally the primary producer of high-quality open-source models.

The group's interpretation: This is deliberate strategy, not generosity. Open-source Chinese models destabilize the commercial moats of US AI companies while creating dependency on Chinese-originated technology. It mirrors the classic loss-leader playbook from manufacturing.

The Disenfranchisement Risk

The more concerning risk isn't the competition between two rational superpowers — it's the disenfranchisement of everyone else. AI gives any motivated individual access to capability that used to require institutional backing. That's democratizing when the individuals are building businesses. It's destabilizing when the individuals are angry and have nothing to lose.

The Public Perception Problem

AI companies created their own backlash. Early messaging was optimized for Silicon Valley investors: "AI will replace jobs, we'll capture the economic value." That messaging reached the general public, and the public heard a threat, not an opportunity. How you talk about what AI does in your product affects public willingness to adopt it.

The B2B Barrier Is Falling

Historically, B2B startups required deep industry experience and institutional connections. AI is compressing that. Someone with strong sales skills can use AI to learn institutional treasury management well enough to pitch credibly. Startups with zero traditional industry expertise are entering B2B markets and competing effectively.

Who Wins

Two profiles emerged:

The fast learner: Someone who can absorb new domains quickly using AI, move into markets they have no background in, and execute before incumbents adapt. Their advantage is speed and adaptability.

The experienced operator: Someone with deep relationships, institutional trust, and contextual understanding that AI can't replicate. Trust and relationships still close deals in B2B, and those take time to build regardless of the tools available.

The consensus: both profiles can win in different contexts. The founders who combine domain credibility with aggressive use of AI tools are likely the best positioned in most markets.

The New SEO: AI Search Optimization

One founder raised the idea of putting your app on AI platforms (MCP servers for ChatGPT, Claude) so that when someone asks the model about your domain, your product surfaces. This is the emerging equivalent of search engine optimization: instead of ranking on Google, you're ranking in the model's responses.

LinkedIn content was flagged as a specific channel — there are reports that ChatGPT pulls heavily from public LinkedIn posts to identify domain experts. Posting substantively about your field may directly influence whether your product appears in AI-generated recommendations.

The Infinite Call Center Thought Experiment

Stop thinking about marketing the old way. Ask yourself: "What if I had an infinite call center that could reach every possible customer individually?" That's no longer hypothetical.

The practical approach:

1. Articulate the problem clearly to a frontier model: "Here is my product. Here is my target customer. Here are the channels available to me. Build me a go-to-market plan with testable approaches."
2. Allocate a small budget per approach (~$1,000) to validate whether it's scalable.
3. Capture data from each test and feed it back into the model for iteration.

Content as a Hook

Before asking for the email, give them something they can't easily get elsewhere: a weekly industry digest, aggregated data, insights from your domain expertise. The landing page has to deliver so much value that giving a real email address feels like a reasonable exchange.