Good AI + Agents UCP

𝗚𝗼𝗼𝗱 𝗔𝗜 + 𝗮𝗴𝗲𝗻𝘁𝘀: Google's new Universal Commerce Protocol lets you unify various agent protocols. But real-life questions need rich and real-life answers (i.e. more product data).

In addition, Google’s new ‘business agents’ can act as your brand and chat with users, directly from within Google Search.

When we change from personally browsing, comparing, clicking to something like agents ‘doing the shopping’ - then the actual shops need to transform as well. Your teams can no longer just supply static product feeds plus support community reviews. The agents want rich and flexible data, made for all kinds of questions.

Here are some of the consequences of Google’s  UCP in my view:

→ 𝗠𝗼𝗿𝗲 𝗖𝗼𝗻𝗻𝗲𝗰𝘁𝗶𝘃𝗶𝘁𝘆: The new UCP acts as the universal translator, allowing customers to use their own AI interface (e.g. Gemini) to brief an agent. Or talk directly to the retailer’s inventory (backend/OMS). And the retailer doesn’t need to do custom integrations for every single type of agent.

→ 𝗠𝗼𝗿𝗲 𝗖𝗼𝗻𝘁𝗲𝘅𝘁: A shop’s marketing and commerce team shifts from keywords to “content & context”. This is fresh data to satisfy AI agents’ new requests. Could be e.g. a product’s compatibility or usage scenario (as in the ‘bike pedal’ example). Then the agent (i.e. customer) can make a decision.

→ 𝗠𝗼𝗿𝗲 𝗡𝗲𝗴𝗼𝘁𝗶𝗮𝘁𝗶𝗼𝗻: The usual promotions won’t be static and rule-based. They will rather be dynamic offers, made for moments of ‘high intent’. They can be injected precisely when e.g. a business agent or shop-AI detects a comparison with your prime competitor.

→ 𝗟𝗲𝘀𝘀 𝗙𝗿𝗶𝗰𝘁𝗶𝗼𝗻: A new type of sales chat (via the ‘business agent’) and checkout travels with the user (e.g. buy via Google Pay directly in the chat). Both make the interaction less ‘visible’ to your original idea of your shop. But both also remove any delays and "link hopping to pay" - which usually reduces conversion.

Should UCP become widely adopted (list of early adoptees include Walmart, Carrefour and Zalando), then commerce becomes more fluid, happening partly inside an AI or agent environment. Two key questions for retailers to discuss now would be:

• 𝗘𝘅𝘁𝗲𝗻𝗱𝗶𝗻𝗴 𝘁𝗵𝗲 𝗱𝗮𝘁𝗮: How do you extend your current data set? Where does this intelligence for ‘high-definition’ feeds reside in the company? How can you allow AI agents to effectively ‘see and discuss’ your products with customers?

• 𝗧𝗲𝘀𝘁𝗶𝗻𝗴 𝗶𝗻 𝗮 𝘀𝗮𝗻𝗱𝗯𝗼𝘅: Before you release any new (and potentially undercooked) business agents, use the new studio tools to internally simulate conversations. Score your business agent’s responses and see how margins, loyalty or brand get affected.

Link to Google resources:
Google's business announcement
'Under The Hood' for developers:

#𝘎𝘰𝘰𝘥𝘈𝘐 #𝘈𝘨𝘦𝘯𝘵𝘪𝘤𝘊𝘰𝘮𝘮𝘦𝘳𝘤𝘦 #𝘎𝘰𝘰𝘨𝘭𝘦𝘜𝘊𝘗