Carve Designs blocks AI agent traffic: a merchant signal worth watching
Carve Designs says it's actively blocking AI agents while Shopify and Salesforce ship LLM integrations. What the split means for eCommerce teams.
Apparel brand Carve Designs is actively blocking AI agent traffic to its store, according to comments from cofounder Thayer Sylvester reported by Digital Commerce 360. The story is short on technical specifics, but the position itself is notable — and it lands the same week that Shopify and Salesforce have been pushing in the opposite direction with LLM integrations.
What happened
In an interview with Digital Commerce 360, Sylvester said Carve Designs is "actively blocking traffic from AI agents" and has invested in technology to do so. She also indicated the approach is not permanent and could change.
The reporting frames this against a broader market shift: Shopify and Salesforce have built integrations with large-language models, opening commerce surfaces to agent-driven shopping flows. Details on exactly which agents Carve is blocking, what tooling it uses, or what triggered the decision were not disclosed in the available summary. Details remain limited at the time of writing.
Background and context
AI agents — autonomous browsing and purchasing flows driven by LLMs from OpenAI, Anthropic, Perplexity, and others — have moved from demos into real traffic patterns over the past year. They show up as scrapers training models, as research assistants summarising product pages, and increasingly as shopping intermediaries that read catalogues and trigger checkout flows on a user's behalf.
Merchants now face a genuine policy question: treat agent traffic as a new acquisition channel, treat it as bot abuse, or split the difference by allowing some agents and blocking others. Tooling vendors including Cloudflare, Fastly, DataDome, and HUMAN have shipped agent-aware controls, and Shopify and Salesforce have moved toward sanctioned LLM integrations rather than leaving the question to merchants alone.
Why it matters for eCommerce teams
Carve's stance is a useful data point because most public commentary has assumed merchants will welcome agent traffic. A direct-to-consumer brand choosing to block — and investing in technology to enforce it — suggests the calculus is more complicated. Concerns likely include catalogue scraping for model training, conversion attribution breaking when an agent sits between the shopper and the storefront, inflated infrastructure cost from non-human traffic, and brand experience risks when an LLM summarises a product page incorrectly.
For Adobe Commerce and Shopify Plus teams, the practical work is the same regardless of which side of the policy you land on. You need to identify agent traffic in your logs, separate it from search crawlers and conventional bots, and decide per-agent whether to allow, throttle, or block. That requires WAF rules tied to user-agent and IP intelligence, structured data that an allowed agent can actually read, and analytics that don't silently fold agent sessions into human conversion rates. Teams running eCommerce technical audits increasingly find that agent traffic is already material — and already mis-classified.
The Shopify and Salesforce integrations matter here because they create a sanctioned path. A merchant on Shopify can plausibly block unsanctioned agents while accepting traffic from agents that route through Shopify's own integration layer, where pricing, inventory, and policy can be enforced.
Key points
- A direct-to-consumer brand is publicly blocking AI agent traffic and has invested in tooling to do so.
- The decision sits against Shopify and Salesforce shipping sanctioned LLM integrations.
- Merchants now need an explicit, per-agent policy — not a default of "allow all bots that aren't obviously malicious."
- Identifying, classifying, and segmenting agent traffic in analytics is a prerequisite to any policy decision.
- The right answer is likely to evolve as agent behaviour, attribution, and platform integrations mature.
Our take
In our view, blocking by default is a defensible short-term posture for brands whose value sits in curation, photography, and brand voice — exactly the assets an LLM flattens when it summarises a PDP. Where we'd push back is on treating the decision as binary or permanent. Agent traffic is not one thing; a training scraper, a research agent, and a checkout-capable agent have different economics, and a single block rule treats them identically.
What we'd recommend for mid-market merchants: instrument first, then decide. Get agent traffic labelled in logs and analytics, quantify its share and its conversion behaviour, and then write a policy per agent class. A blanket block adopted without measurement is just as fragile as a blanket allow.
What to watch next
Watch whether more brands go on record with their agent policies, and whether Shopify and Salesforce expose merchant-side controls that make "allow sanctioned, block everything else" a one-click setting. If platform-level controls arrive, the cost of holding a strict position drops sharply, and the conversation shifts from infrastructure to merchandising.
Source: Based on reporting from Digital Commerce 360, published on April 30, 2026.
Disclaimer: This article is based on information available from the cited source at the time of writing. MagentoInfo Corp added independent context and analysis. Details may change as the story develops.