barter.computer

Personal agents like Muse are being distributed at massive scale. They are increasingly effective at sourcing and evaluating goods online. They can take a transaction right to the point of sale. Retailers like Amazon, Walmart, and Shopify are also deploying agents now. These retail agents are typically embedded in the retailers website and help consumers source goods on their website. Retail agents have the issue that they can only see their own product. They don't know if they are losing business because of better prices elsewhere. A person leaving the conversation could either have found a better alternative, thought it was too expensive, or even just forgotten and the vendor then has no control over that sale. Personal agents face a different problem. ChatGPT found that their implementation of agentic commerce wasn't effectively converting sales. People enjoyed agentic search and discovery but checkout they preferred to do themselves. This is what the barter.computer protocol aims to solve. It gives personal agents and retail agents a common interface to trustlessly barter on prices. A personal agent goes and sources a product, barters with retail agents and produces a price lower than the best found publicly available price. Now this deal exists for this agentic session, not publicly. A user goes and buys that product through their agent. The protocol here demonstrates this agentic bartering concept on retail products because of the simplicity of finding online pricing. As more companies move to support agentic traffic, prices will be queryable for not just goods but real-world services. General purpose models like Meta's mean that same protocol can be applied to sourcing a plumber using photos, finding car insurance based on personal history, and beyond. I think that this style of agentic barter will proliferate quickly once one big vendor implements an agent that accepts barter and consequently drives significant volume through agentic transactions.