On September 29, 2025, OpenAI enabled in-chat purchases through ChatGPT for U.S. consumers. Within weeks, brands were rebuilding their sales copy for a buyer that does not have eyes. The Seattle Times reports the rewrite is now in full swing.

The paper's headline — "As AI agents begin shopping, brands are changing their sales pitch" — captures the inflection in a single sentence.

What does "begin" tell us about timing?

The verb signals an early, active shift. "Changing their sales pitch" implies an ongoing migration rather than a finished overhaul. For any sales leader, that puts the question on this quarter's roadmap rather than next year's review.

What changes when the shopper is a model?

Three things, according to the pattern consumer brands describe.

Structure, then specs, then proof. Schema markup, JSON-LD product feeds, GTINs, and machine-readable spec sheets become the actual sales pitch. Models cannot be persuaded by lifestyle photography; they read fields. A SKU without a manufacturer part number is invisible to an agent.

Descriptions that used to open with brand narrative now open with the facts an agent compares: dimensions, weight, materials, voltage, compatibility, country of origin.

Reviews surface as structured fields — rating, count, verified-buyer flag — before any prose summary.

Why did this accelerate in late 2025?

The shift tracked a cluster of agentic-commerce launches through late 2025.

ChatGPT processed its first in-chat purchases in September through partnerships with Stripe, Etsy, and Shopify. Google expanded AI Mode in search results. Walmart deployed Sparky across its main app. Amazon's Rufus handled a growing share of pre-purchase queries.

Each release moved the consumer one step further from the brand's website — and one step closer to an intermediary that reads rather than browses.

What does a bot-friendly sales pitch look like?

The brands furthest along describe four practical shifts in their copy:

• Product descriptions open with dimensions, weight, materials, voltage, and compatibility.
• Pricing appears in plain numerals with currency code and tax-inclusive status.
• Reviews surface as structured fields before any prose summary.
• FAQs answer the model's likely follow-up questions in schema-tagged blocks.

Which tools are driving the migration?

Four categories of vendors now sit between the catalog and the agent.

Feed-management platforms like Feedonomics, DataFeedWatch, and GoDataFeed auto-generate agent-readable schemas from existing product data. Marketplace connectors from Shopify and BigCommerce ship pre-built feeds tuned for agent consumption. Agent-side analytics tools such as Profound, Peec AI, and Otterly show brands which queries their products surface in.

The tooling means the rewrite is not a copywriter's project — it is a marketing-ops deployment owned by the team that already runs the data layer.

What does this mean for sales teams?

For a B2C sales organization, the consequence is structural. The buyer of record is now a language model acting on a consumer's behalf. Brief templates that used to gate creative on brand fit now gate it on whether a product feed can pass an agent's parser. Marketing operations — not copywriters — increasingly own that step.

Compensation plans that rewarded traffic-based conversion need new metrics: agent citations, feed health scores, and the share of orders where the agent's first-choice product matched the merchant.

What comes next?

The next threshold is agent-initiated negotiation. Comparison tools today return ranked lists. Tools in 2026 will request quotes, apply loyalty terms, and bundle complementary SKUs in real time. The brands that ship agent-readable catalogs this quarter will be the ones whose pricing logic the agents learn to invoke by default.