Agentic Commerce Commons (ACC) at MIT

Agentic Commerce
Commons (ACC) at MIT

Retail at a Crossroads: The rules of agent-driven commerce are being written now.
Massachusetts Institute of Technology
MIT Media Lab (E14), 6th floor  ·  Cambridge, Massachusetts
The Challenge
In the emerging AI architecture for retail, a single AI provider has a conflict of interest, both advising the buyer and gathering revenue from product placement for the seller. The AI intermediates nearly all interactions between buyer and brand.

There is a single merged catalog representing all products from every seller, and the product discovery process never needs to touch a retailer site. Under these conditions a retailer's remaining surface area is price and fulfillment. Brand, assortment, merchandising judgment, and customer relationship are owned by the AI platform.

The Opportunity
Buyer Agents who represent consumers connect securely and privately with Seller Agents who represent retailers, with networks and service providers supplying identity, provenance, and payments.
The Default
Today’sAI AgentsBUYERSSELLERS
The Vision
The connection happens in an open discovery network.
Networks and service providers for identity, provenance, payments, and other services.
Participants

This workshop will bring together retail and brand CEOs, commerce platforms, card networks and issuers, payment platforms, researchers, builders, investors, and infrastructure providers to examine and design an open market system that directly connects on-device AIs loyal to each buyer with cloud AIs loyal to each seller.

Outcome

Retailer and brand inputs and expert conversations summarized. Builder working groups formed. We take the principles, draft standard, and demo system to the National Retail Federation (NRF), Javits Center, NYC, January 10–12.

When & Where

Friday, December 4, 2026, at the MIT Media Lab (E14), 6th floor, Cambridge, Massachusetts. 9am–7:30pm including dinner. Agenda and registration to follow.

Foundation
and Research

An open source foundation provides protocols and community guidelines for the open discovery network, with academic work contributing market mechanisms, measurement, and transparency so that the market can operate towards its stated principles.

An agent market
that brings together and extends existing protocols including UCP, ACP, AP2, and A2A.
A not-for-profit foundation
to steward mechanism design, measurement, and governance.
A shared simulation environment
based on the Trading Agent Competition (Wellman et al).
Design principles
that resist formation of monopolies or monopsonies.
A whitepaper describing the proposed market design is available on request.
Working Group
Topic Areas
Mechanism design, monitoring & governance

The concept of a standard is evolving. Agents are capable of re-writing/extending protocols and changing behavior on the fly. Agents can also choose any market within which to participate, and rapidly switch. In the Darwinian selection of behaviors and markets, how do we design and incentivize a race to the top, rather than a race to the bottom?

01 Discovery, recommendation, & advertising
02 Price negotiation & auction design
03 Dispute resolution & escrow
04 Data & privacy
05 Identity, trust, & provenance
Academic
Organizers

Convened by Ben Vigoda (MIT Design Intelligence Lab), Negin Golrezaei (MIT Sloan), and Vivek Farias (MIT Sloan).

Ben Vigoda
Negin (Nikki) Golrezaei
Vivek Farias
Yiling Chen
Michael Wellman
To be announced
SAVE
THE
DATE
Friday, December 4, 2026
Massachusetts Institute of Technology
MIT Media Lab (E14), 6th floor  ·  Cambridge, Massachusetts
9am–7:30pm including dinner
Agenda and registration to follow. To speak, sponsor, join a working group, or nominate a colleague, reply to AJ Abraham, workshop coordinator, aj.abraham@agenticcommercecommons.org.

Download the save-the-date (PDF)