A retailer drawing its own line on agent identification
Home Depot CIO Angie Brown has staked out a specific requirement for AI shopping agents operating on the retailer's site: they must identify themselves as agents and be clearly distinguishable from ordinary bot traffic before Home Depot will let them complete a transaction on a customer's behalf. That requirement puts a concrete operational bar on a problem most of the industry has so far discussed only in general terms.
The timing matters. Agentic commerce has moved quickly from a conceptual demo to live infrastructure at major retailers over the past year, and Brown's comments are among the clearest statements yet from a retail CIO about what technical and procedural conditions a retailer actually wants satisfied before granting an AI agent the same purchasing access as a logged in customer.
Where Home Depot sits relative to the rest of the industry
Home Depot's position is notably stricter than several major peers. Walmart, Best Buy, Sephora, and Shopify have each signaled more openness to agentic traffic reaching their checkout flows, while Amazon has taken the opposite extreme, blocking Meta's Muse agent outright over what Amazon considers insufficient bot identification. Home Depot's stance lands between those two poles: not blocking agents outright, but requiring a specific identification standard before allowing them through.
That middle position is arguably the more operationally demanding one. Outright blocking requires no ongoing technical cooperation with agent platforms, and outright openness requires no enforcement mechanism. Requiring verifiable identification means Home Depot has to build and maintain the actual verification infrastructure, and it has to do so in active coordination with every agent platform it wants to allow through.
The live test case: Home Depot's talks with Meta
Home Depot is reportedly in active discussion with Meta about exactly how its Muse shopping agent should present itself to Home Depot's systems at checkout, and Meta has been developing self-identification capability for Muse that appears directly responsive to this kind of retailer requirement. That makes the Home Depot-Meta relationship a concrete, trackable test of whether agent identification standards can actually be negotiated bilaterally between a retailer and a platform.
If this negotiation produces a working identification approach, it becomes a template other retailers can reference rather than building from scratch, and other agent platforms beyond Meta would have a concrete precedent to measure their own agents against. If it stalls or produces something retailers consider inadequate, that failure is equally informative about how hard this problem actually is to solve through bilateral deals rather than an industry wide standard.
Why there is no shared standard yet, and what that costs retailers
No industry wide technical standard currently governs how an AI shopping agent should identify itself to a retailer's systems, which means every retailer, Home Depot included, is effectively negotiating its own bar with each agent platform independently. That fragmentation is expensive: a retailer has to build verification logic that may need to differ by platform, and an agent platform has to adapt its identification approach differently for every retailer it wants to transact with.
This is the kind of problem that typically gets resolved either through an emerging de facto standard set by whichever early mover's approach other players converge on, or through a formal industry standards effort once enough friction accumulates. Home Depot's specific, publicly stated requirement is exactly the kind of concrete reference point that could anchor an informal standard before a formal one exists.
What this means for retail CIOs building their own agent policy
Retail technology leaders who have not yet set an explicit policy on agentic traffic should treat Home Depot's public stance as a useful starting template rather than reinventing the identification requirement from first principles. The core question Brown's position answers, agents must identify themselves and be distinguishable from bots, is a reasonable minimum bar that balances allowing legitimate agentic commerce against the fraud and abuse risk of indistinguishable bot traffic.
Retailers should also watch the outcome of Home Depot's Meta negotiation closely, since whatever technical mechanism emerges from it, whether that is a verifiable agent credential, a specific header or protocol signal, or something else, will likely become the first concrete template available for retailers who want a specific requirement to point agent platforms toward rather than negotiating abstract principles from scratch.
The open question that will determine how this settles
The central unresolved question is whether agent identification converges toward one dominant technical approach that most retailers and platforms adopt, or whether it stays fragmented with each retailer-platform pair negotiating its own bilateral arrangement indefinitely. Home Depot's early, explicit stance makes it one of the more influential data points in which direction that convergence goes.
Retail technology leaders should track whether Amazon's outright block of Muse softens once Meta's self-identification capability matures, since Amazon moving toward conditional acceptance would be a strong signal that a workable identification standard has emerged, while continued blocking would suggest the technical or trust bar for agent identification remains unresolved even for platforms actively building toward it.
