We're at an inflection point in how goods and services are bought and sold online. The rise of capable AI agents isn't just adding a new feature to existing commerce — it's creating an entirely new paradigm.
The numbers
Research from multiple analyst firms converges on a striking projection: by 2028, approximately 40% of online purchases will be initiated, researched, or completed by AI agents acting on behalf of consumers. That's not a typo. Within two years, nearly half of all e-commerce transactions will have an AI agent somewhere in the decision chain.
This isn't speculation based on hypothetical technology. The building blocks are already in place. Consumer AI assistants from major tech companies already have hundreds of millions of active users. These assistants are rapidly gaining the ability to take actions — not just answer questions — and purchasing is the highest-value action most consumers want delegated.
What changes for sellers
The shift to agent-mediated commerce changes the seller's optimization target in fundamental ways.
Discovery shifts from visual to data-driven. On a traditional marketplace, discovery is driven by search rankings, sponsored placements, and visual appeal. In an agent-mediated world, discovery is driven by structured data quality, price competitiveness, and fulfillment reliability. The seller's "storefront" isn't a webpage — it's a data schema.
Brand matters differently. Agents don't have emotional responses to brands. They evaluate brand reputation as a signal — through review aggregation, return rates, and consistency scores — but they don't feel brand loyalty. This levels the playing field for smaller sellers with excellent products and creates pressure on established brands that rely more on perception than substance.
Customer relationships become indirect. When an agent buys on behalf of a consumer, the seller's relationship is with the agent ecosystem, not the end buyer. This creates new dynamics around loyalty, retention, and customer service. Sellers who build excellent post-purchase experiences — fast shipping, proactive communication, easy returns — will earn higher agent ratings and more repeat recommendations.
What changes for platforms
Existing marketplace platforms face a strategic choice: adapt for agent commerce or risk being disintermediated.
Platforms that expose structured, machine-readable product data through APIs will become the preferred shopping destinations for AI agents. Platforms that rely on proprietary search algorithms, walled-garden ecosystems, and human-centric UX will find their traffic shifting to agent-friendly alternatives.
This is why we built Nohi as an agent-first platform from day one. Every design decision — from our product schema to our checkout flow to our review system — prioritizes machine readability and programmatic access.
What changes for consumers
For consumers, agent-mediated commerce promises significant improvements in three areas:
Time savings. The average online purchase involves 20-30 minutes of research, comparison, and checkout. An AI agent can compress this to seconds. For routine purchases — groceries, household supplies, subscriptions — the agent can handle everything autonomously, freeing up hours per week.
Better decisions. AI agents can evaluate far more options than a human shopper. They can compare prices across every available marketplace, cross-reference reviews from multiple sources, and factor in personal preferences like sustainability commitments or dietary restrictions — all instantaneously.
Reduced cognitive load. Decision fatigue is real. By delegating purchasing decisions to a trusted agent, consumers can reserve their mental energy for decisions that actually require human judgment — creative choices, relationship decisions, career moves — rather than spending it on which brand of dish soap to buy.
The trust question
The critical enabler for all of this is trust. Consumers need to trust that their AI agent is acting in their interest, not the seller's. Agents need to trust that product data is accurate and pricing is fair. Sellers need to trust that the platform will deliver qualified, high-intent traffic.
At Nohi, we're building trust through transparency. Our verification systems, structured review formats, and open pricing policies are all designed to create an environment where trust can be established programmatically — not just through brand perception or marketing claims.
The timeline
We believe the transition to agent-mediated commerce will unfold in three phases:
Phase 1 (2025-2026): Assisted browsing. AI agents help consumers research products and compare options, but the human makes the final purchase decision. This is where we are today.
Phase 2 (2027-2028): Delegated purchasing. Consumers set budgets, preferences, and rules, and AI agents execute purchases autonomously within those constraints. This is where the 40% projection comes in.
Phase 3 (2029+): Predictive commerce. AI agents anticipate needs based on usage patterns, calendar events, and life changes, proactively purchasing items before the consumer even thinks to ask. This phase will require deeper integration with personal data and significantly higher trust thresholds.
We're building Nohi to be the infrastructure layer that supports all three phases. The marketplace that agents trust today will be the marketplace that agents rely on tomorrow.
The future of commerce is autonomous. The question isn't whether it will happen, but who will build the infrastructure to support it. We intend to be that infrastructure.
