On 6 July 2026, Salesforce made Agentforce Commerce generally available, and it did so with unusually good timing: right before peak season, when every retailer’s revenue is decided in a few frantic weeks. The headline is that Salesforce’s Shopper, Buyer, and Merchant agents are now live, and that they no longer live only on your own storefront. They are wired natively into ChatGPT, into Google Search including AI Mode, with the Gemini app following this summer. In plain terms, the AI agent that sells your products can now do so inside the places your customers already spend their attention. That is one of the most genuinely exciting shifts in commerce this decade.
The numbers Salesforce shared during testing are the kind that make revenue leaders sit up. Retailers running their own shopper agents grew sales 59 percent faster than those still on the sidelines, and AI-referred traffic converted at eight times the rate of social. Whatever discount you apply to vendor figures, the direction is unmistakable: agentic commerce is not a future experiment, it is a channel that is opening right now, and it is still largely uncontested. The teams that move first get to shape how their brand shows up in an AI-mediated shopping experience before their competitors even realize the channel exists.
There is a catch, and it is a productive one. An agent selling on your behalf is only ever as good as the commerce data sitting behind it. The moment an agent quotes a price, confirms stock, or promises a delivery date inside ChatGPT, your catalog, inventory, and pricing data stop being an internal back-office concern and become the customer-facing product itself. That is not a reason to wait. It is a reason to get your data house in order now, while the channel is still young and the stakes of a mistake are low.
What did Salesforce actually launch on 6 July 2026?
What did Salesforce launch in the July 2026 Agentforce Commerce release? Salesforce made three commerce agents generally available: Shopper Agent for B2C, Buyer Agent for B2B, and Merchant Agent for back-office operations. The release moves each one past the bolt-on chatbot model and wires it natively into your catalog, inventory, order management, and customer data from day one. The most consequential detail is distribution: the agents are natively integrated with ChatGPT and Google Search, including AI Mode, with the Gemini app arriving this summer, so an agent can transact for you inside those surfaces rather than only on your own site.
Salesforce called this its biggest Agentforce Commerce release yet, and the timing ahead of peak season is deliberate. This is the quarter when the platform wants retailers to test agentic selling against real demand rather than in a sandbox. For anyone who has run a Commerce Cloud storefront, the significance is that the agent is not a widget bolted onto the checkout. It is connected to the same systems of record your human team uses, which is exactly what makes the opportunity real and the data readiness question unavoidable.
What is agentic commerce?
What is agentic commerce? Agentic commerce is a model in which an AI agent, rather than a person clicking through a website, carries a shopping or buying task from start to finish: it discovers products, answers questions, checks live inventory, applies the right pricing, and completes the purchase, often inside a conversation on a third-party surface like ChatGPT or Google. The customer states an intent in natural language, and the agent does the work of turning that intent into a completed order. It compresses the old browse-compare-checkout journey into a single exchange.
Why this matters for CRM and commerce leaders is that the agent, not the customer, is now the one reading your data. A shopper skims a product page and forgives a small inconsistency. An agent takes your feed literally: if two systems disagree on price or stock, the agent will confidently state one of them, and it may be the wrong one. The shift from human interpretation to machine execution is what raises the bar on data quality, and it is also what makes the upside so large for the teams whose data is clean enough to trust.
What do the Shopper, Buyer, and Merchant agents actually do?
What do the Agentforce Commerce agents do? Each agent owns a distinct part of the commerce workflow. The Shopper Agent handles B2C: it carries the customer conversation from discovery through checkout and into service, speaking in your brand voice, checking live inventory, confirming carrier cut-offs, offering store pickup, and closing the sale in one conversation. The Buyer Agent handles B2B: it meets buyers inside WhatsApp and SMS, supports SKU confirmation from an image, applies contract pricing, and completes orders without a portal login. The Merchant Agent runs the back office, letting your team organize catalogs, sort products, and respond to trends in plain language.
Read those capabilities closely and a pattern jumps out: every one of them leans on a specific slice of your data being correct in real time. Live inventory, carrier cut-offs, store availability, contract pricing, image-matched SKUs. These are not marketing features so much as data dependencies. The Shopper Agent that confirms a carrier cut-off is only as reliable as your fulfillment data; the Buyer Agent that applies contract pricing is only as trustworthy as your pricing records. The agents are impressive precisely because they act on this data directly, which is the whole reason getting it right beforehand pays off so quickly.
Why does selling inside ChatGPT raise the stakes on your data?
Why does agentic commerce raise the stakes on data quality? Because the error is now public, immediate, and transactional. When your catalog was something a shopper browsed on your own site, a stale price or a mislabeled product was an internal problem you could quietly fix. When an agent quotes that same price inside ChatGPT and a customer acts on it, the mistake becomes a promise made in your brand’s name on a platform you do not control. Agentic commerce turns every data inconsistency from a private inconvenience into a customer-facing commitment.
This is where the good news and the discipline meet. The same native connection that makes these agents so powerful, the fact that they read straight from your catalog, inventory, and order systems, means there is no editorial layer between your data and the customer. There is no merchandiser to catch the odd product before it ships to a shopper’s screen. That is not an argument against agentic commerce; it is an argument for treating your product, pricing, and inventory data as the customer experience it has now become. Teams that make that mental shift tend to find the work is more contained than they feared, and the payoff faster than they expected.
Is your commerce data ready for agents?
How do you know if your commerce data is ready for AI agents? Ask four questions. First, is there a single source of truth for price and inventory, or do your storefront, ERP, and order system sometimes disagree? Second, is your product data complete and structured, with the attributes, images, and descriptions an agent needs to answer a question accurately? Third, is fulfillment data, carrier cut-offs, store stock, delivery windows, current enough for an agent to make a real-time promise? Fourth, for B2B, is contract pricing modeled cleanly enough that an agent can apply the right price to the right account without a human check?
If any of those answers is uncomfortable, that is useful information, not a verdict. It simply tells you where to point your effort before you let an agent transact at scale. The encouraging reality is that most organizations already hold this data; it is just scattered across systems that were never asked to agree with each other in real time. Reconciling them is a well-understood exercise, and doing it now, before peak season and before the channel matures, is far cheaper than doing it after an agent has confidently quoted the wrong number to a thousand shoppers.
Where should commerce leaders start?
What should commerce leaders do first about Agentforce Commerce? Start with a contained pilot on a well-understood product range rather than your entire catalog. Pick a category where your price, inventory, and fulfillment data are already trustworthy, stand up a Shopper or Buyer Agent against it, and watch how the agent behaves with real customers before widening the scope. Use the pilot to expose the data gaps that only surface under live conditions, then fix those before you extend the agent to messier parts of the catalog.
In parallel, decide who owns the data the agents depend on. Agentic commerce blurs the old line between the merchandising team that curates the catalog and the operations team that maintains the systems of record, because now both feed the same customer-facing agent. Naming a clear owner for product, pricing, and inventory accuracy, and giving them the mandate to enforce a single source of truth, is the least glamorous and most valuable move you can make this quarter. It is also the step that turns a promising pilot into a channel you can scale with confidence.
The Sirocco perspective
In our work across Salesforce, HubSpot, and Dynamics 365, the pattern we expect with Agentforce Commerce is the one we see with every powerful new capability: the technology will work, and the results will come down to whether the data underneath it was ready. The clients who win with agentic commerce will not be the ones who deployed fastest. They will be the ones who treated their product, pricing, and inventory data as a customer-facing asset before they pointed an agent at it. As an independent CRM and commerce partner, we are not tied to one vendor’s account of how ready your estate is. That lets us give you a straight read on where your data can be trusted to sell on its own and where it cannot, yet.
Agentic commerce is one of the clearest opportunities in the space right now, and the path to capturing it is more practical than the hype suggests. Get the data foundations right, pilot on ground you trust, and name an owner for accuracy, and you can move into this channel early without betting your peak season on it. If you would like an independent view on whether your commerce data is ready for agents, schedule a consultation with our team.
Get in Touch
If Agentforce Commerce is on your roadmap and you are not yet sure your catalog, pricing, and inventory data are ready to sell on their own, that is exactly the question worth working through before peak season. Tell us where your commerce data lives today and we will help you map a sensible path to your first agent.
