The Connective Tissue Argument: Why Agentic Commerce Needs a Neutral Marketplace

I want to make an argument tonight that I think is unfashionable in this room, and I want to make it directly so we can discuss it.

The argument is this: every closed-platform play in agentic commerce that is currently being built: the brand consortia, the carrier-led customer-experience platforms, the bank-led financial advocates, the retailer-led shopping agents, will lose. They will lose for the same reason walled-garden directories lost to the open web in the 1990s. They will lose for the same reason proprietary instant-messaging networks lost to SMS and then to internet protocols. They will lose because the topology of value in agent-to-agent commerce is not aligned with closed-platform economics.

The companies that win the next decade in this space will be the neutral marketplace plays. Not because their teams are smarter or their fundraising is better or their technology is more sophisticated. Because the topology is what determines who wins, and the topology favors the layer that is the connective tissue between every consumer agent and every brand agent.

Let me unpack that.

Why brands cannot build the consumer agent

Every Fortune 500 retailer in this room has, at some point in the last two years, asked a version of the same question: should we build our own consumer agent? Some of you have already started. The pitch is intuitive. You have the customer relationship. You have the data. You have the brand trust. Why would you cede the consumer-side agent to a third party?

Here is the structural problem. A consumer's agent is only useful to the consumer if it represents the consumer across every brand the consumer transacts with. The consumer has fifty to two hundred recurring vendor relationships. An agent that represents only one brand is not an advocate; it is a chatbot in agent-shaped clothing. From the consumer's perspective, the value of the agent is exactly proportional to how much of their brand-relationship surface it covers, and the only way to cover all of it is to be a third party that is not affiliated with any single brand.

A bank that builds its customers a financial advocate that only handles the bank's products solves a small slice of the consumer's problem. The advocate is not actually advocating; it is selling. A retailer that builds its customers a shopping agent that only operates within the retailer's catalog is not helping the customer find the best deal; it is helping the customer find the best deal within the retailer's inventory. The structural asymmetry of who the agent represents, the consumer or the brand, is not solved by the brand calling its agent a "consumer advocate." It is determined by who pays for the agent and whose interests the agent's loyalty runs to.

This is why every brand-led consumer agent eventually plateaus at the boundary of the brand's own footprint. The consumer's agent must be a third party, by structural necessity, not by a strategic choice the brand could make differently.

Why brands also cannot build the marketplace

The next pitch you hear is: fine, we will not build the consumer agent, but we will build the marketplace where consumer agents and brand agents meet. We have the relationships, we have the channel power, we have the infrastructure.

This pitch fails for a different reason. A two-sided marketplace has to be neutral to attract both sides. A consortium of banks cannot host a marketplace that competing banks will join. A retailer-led commerce platform cannot host a marketplace that competing retailers will trust. Every brand that joins a brand-led marketplace is implicitly sending high-intent customer signal to a competitor's infrastructure. The economics of the marketplace are the inverse of the economics of the brand: the marketplace wants to aggregate every brand, the brand does not want to participate in any aggregation it does not control.

This is not a temporary friction that better governance can solve. It is a structural feature of two-sided networks that has played out the same way in every market they have appeared in. The marketplace must be neutral, because non-neutrality reduces the participation rate, and a marketplace with low participation has no value to either side.

What the connective tissue actually does

The neutral marketplace layer in agentic commerce, the connective tissue: has three functions, none of which any brand-led play can perform structurally.

It standardizes the protocol grammar between agents. Every consumer agent talks to every brand agent in the same wire format, with the same identity claims, the same authority declarations, the same receipt schemas. The marketplace defines the standard because it is the only party with no incentive to favor one side over the other.

It aggregates trust signals. The reputation of a brand agent, its honored-commitments rate, its dispute outcomes, its consumer-satisfaction signal, is measured across every consumer agent that has ever transacted with it. The reputation is portable across the marketplace and credible because the marketplace has no stake in any specific brand's outcome. This trust-aggregation function cannot be performed by a brand-led platform without conflict-of-interest concerns the brand cannot defuse.

It generates the negotiation corpus. Every transaction that runs through the marketplace produces a signed trace. Aggregated across millions of transactions, these traces become the training data for the next generation of agents on both sides. The corpus is proprietary to the marketplace and inaccessible to competitors. This is the data-moat that compounds with scale and is structurally unavailable to any single brand.

These three functions: protocol standardization, trust aggregation, negotiation-corpus generation, are what make the marketplace layer the locus of long-term value in agentic commerce. They cannot be performed by the brand-led plays. They will accrue to the neutral layer by structural necessity.

What this means for the brands in this room

I am not arguing that brands should be passive in this transition. I am arguing that the brands that win will be the ones that figure out their role in the new topology rather than the ones that try to recreate the old topology with new technology.

Three actions matter.

First, build a brand-side agent that is genuinely good at representing your brand to a consumer's agent. That is the role you can play, and it is a real role. A great brand agent: one with cleanly bounded authority, fast resolution times, transparent receipts, and a high honored-commitments rate, will have customers' agents preferring to negotiate with it. That preference compounds into customer retention in a market where the consumer's loyalty has shifted from the brand to the consumer's agent.

Second, integrate with the marketplace layer early. The marketplace will eventually have the customer's attention through their agent. Brands that integrate early will be the brands the marketplace surfaces; brands that integrate late will be the brands the customer's agent does not consider. This is the same dynamic as early Google search-engine optimization: there was a window where being indexable mattered, and the brands that ignored it for too long lost the channel.

Third, do not invest in building a competing marketplace unless you are prepared to be a structurally non-participatory marketplace. The economics of two-sided networks will not favor your closed play. You will spend a meaningful amount of capital trying to recreate a topology that your competitors will refuse to participate in.

Closing

The argument I want to leave you with: the connective tissue layer in agent-to-agent commerce is not optional. The marketplace will be built. The question is whether your brand will be on it as a strong participant, or off it as a holdout, or trying to be it as a structurally non-viable play.

The brands that get this right will be the brands that survive the transition. The ones that don't will discover that their customers' agents have stopped routing requests to them, and they will not understand why. The reason will be that they tried to own the connective tissue and ended up being routed around by it.

Thank you. I look forward to the discussion.