Better Should Win

For most of the last century, friction has been subsidizing mediocrity.

Businesses survive not because they're the best, but because finding the best is hard. Customers default to convenience, familiarity, proximity, whoever showed up first. Information asymmetry protects the average. Switching costs protect the complacent. The institutions that shape daily life weren't designed to be the best possible version of themselves — they were designed to be functional within the constraints of their era, and many are no longer functional even within those.

Friction is not neutral. It is a tax that consumers and citizens pay so that mediocre organizations can continue to exist.

AI is removing that tax faster than most people have internalized. When agent-to-agent commerce becomes the primary mechanism for B2B transactions, the businesses that survive will be the ones that are genuinely, measurably, demonstrably better. Not better marketed. Not better connected. Better. The same dynamic applies to healthcare, education, public services, every system where friction has been masking failure.

This is the bet behind everything I build.

The transition from friction-subsidized commerce to merit-decided commerce needs infrastructure that doesn't exist yet.

When AI agents start making procurement decisions on behalf of buyers, those agents need ground truth about which sellers are actually better — not which ones tell the best story. That ground truth has to come from somewhere. It has to be built from real conversations, real outcomes, real decisions under real stakes. The corpus required to model commercial trust at this resolution doesn't exist in the public record. Someone has to generate it.

That's what FBED Labs does. Three application surfaces, one neural network, a thesis being tested in real deals.

ConversationOS — the relationship layer. Twelve years of evidence that the best B2B pipelines are built one conversation at a time, and that the right system makes that scale.

ICPX — the prospecting layer. Filters for real humans doing real work, not synthetic engagement.

CASCADE — the conversation layer. Models what actually happens inside commercial conversations, treating selling as a prediction problem rather than a volume one.

The three surfaces feed one model. Every conversation sharpens the substrate. Every outcome labels another data point. The system gets better with use, and the moat is the corpus.

I think the trust layer for agent-to-agent commerce will be one of the most leveraged pieces of infrastructure built in the next decade. I think most of it gets built in the next four years. I think the organizations that own outcome-labeled commercial conversation data at scale will sit in a position the rest of the market can't reach.

That's what we're building toward.

The same thesis applies outside commerce. Operators who are genuinely better at what they do, building in domains where friction has been protecting the incumbents, deserve capital and attention.

Sarnie Social — Ticketed social dinners, no reservations, the sandwich taken seriously. Seif is rebuilding what the British high street should feel like. I think this is what eating out looks like next.

Tipple — Group sport prediction with friends, leaderboards, weekend games. Social sport engineering pulled away from the canonical betting markets. The thing your group chat already does, given proper infrastructure.

Both bets express the same conviction: in the markets that matter to people, better should win, and the operators who build genuinely better experiences will take disproportionate share when the friction protecting the incumbents finally gives way.

The period between now and the merit-decided future is the dangerous part.

When friction collapses, the first thing that happens is not improvement. It is destruction. The organizations and roles that were being subsidized do not gracefully adapt. They fail. And when many of them fail in the same window, the social and economic consequences are real.

I don't think the answer is to slow the transition. I don't think we can. I think the answer is to build the infrastructure, the businesses, and the operators that are genuinely worthy of the future that's coming, so that what survives is worth surviving.

That's what this is about.