LoginSign Up
← Back to E-comm
Weekly · E-comm · Week of July 6, 2026

Commerce platforms race to make themselves agent-drivable

mcpai-agentsfulfillmentforecastingplatform-infrastructurecontent-vs-product
O
By Onur Öztürk
LinkedIn ↗
Generated 1h agoDrawn from 11 products

The week in ecommerce

The clearest pattern this week is commerce infrastructure turning itself into something an AI agent can drive. Wheelhouse shipped an MCP server exposing 58 tools at 1:1 parity with its UI, ShipHero opened its fulfillment data to any MCP client through an AI Toolkit, and Spree Commerce landed 5.5 with a fully-typed Admin API and 25 installable agent skills. These are three different companies — revenue management, 3PL fulfillment, open-source storefront — arriving at the same conclusion in the same seven days: the product surface that matters next is the one a coding agent or assistant can query and operate without a human clicking through screens. This is a step past "we have an API." It is exposing the whole platform, read and often write, to the assistant ecosystem as a first-class interface.

The second pattern is quieter and worth watching against the first: the platforms with the most raw activity are shipping plumbing, not direction. Shopify pushed eleven improvements with zero sparks — metafields as analytics dimensions, app-pixel audit logs, region-specific tax validation — the granular data-model and compliance work that keeps larger merchants on the platform but changes nothing about what Shopify is. Much of the sector's apparent motion is also content, not code: Printful, LoyaltyLion, Brightpearl, Cin7, ShipBob, ShipHawk, Okendo, and Payhip all posted at a steady clip this week with no product signal at all, their velocity carried entirely by SEO and marketing blogs. The real releases are concentrated in a handful of names.

Leaders

Wheelhouse is the sharpest move of the week. Its MCP server exposes 58 tools at full UI parity, so Claude, Cursor, and VS Code can run any pricing, reservation, or comp workflow by prompt, sitting atop a fast-expanding RM API and new neighborhood-occupancy metrics. This is the capstone of a deliberate pivot from a pricing UI to an API-first, AI-drivable platform, with a July hackathon to seed third-party builders.

ShipHero paired one spark with eight improvements, the highest sustained shipping cadence with real product signal in the sector. The spark is an AI Toolkit — a read-only MCP connector plus a Public API Skill — that lets operators point their own agent at orders, inventory, and shipments in plain language. The rest is pack-station ergonomics: tote scanning, order attachments while packing, CSV location edits. Two tracks, both moving.

Spree Commerce shipped 5.5, the one concrete release under a wall of marketing posts elaborating it. The typed Admin API, TypeScript SDK, and 25 agent skills reposition the open-source storefront as an agent-buildable commerce backend. It is the same agent-native bet as Wheelhouse and ShipHero, made from the self-hosted, developer-owned end of the market.

Shiprocket is expanding outward from shipping into an AI product portfolio: this week's AITLAS launch is an AI brand-visibility platform aimed at a world where commercial queries get answered inside assistants, alongside its Address Intelligence geocoding work. The velocity score is inflated by heavy logistics SEO content, but the AITLAS spark and the geocoding improvement are genuine product moves using its merchant base as the distribution wedge.

Katana posted two real improvements led by AI replenishment that forecasts demand twelve months ahead, plus custom fields on sales orders. It is the manufacturing-ERP entry in this week's AI-forecasting thread — smaller in profile than the platform players, but shipping demand-planning capability rather than writing about it.

Wildcards

Shopify is the odd one out precisely because it is the largest: eleven improvements, no spark, no directional shift. The metafield-as-dimension and audit-log work is unglamorous retention plumbing for regulated, multi-location merchants — important, but a deliberate absence of the agent-interface bet everyone smaller is racing toward.

SpotOn surfaces from the restaurant-POS edge of the sector with six improvements in a monthly digest: phone-ordering and table-fill add-ons with no commissions, penny rounding, printing and tip tools for staff. It is off-pattern — operational back-office polish for hospitality operators, not commerce infrastructure — and its low velocity score understates that real, if narrow, shipping is happening.

Themes that compounded

  • MCP and agent-drivable interfaces became the week's dominant direction, with Wheelhouse, ShipHero, and Spree all shipping agent-facing surfaces.
  • AI-assisted forecasting and discovery moved from marketing copy to shipped capability at Katana (replenishment) and Shiprocket (AITLAS).
  • The highest-velocity feeds were mostly content, not code — eight sector products posted zero product signal behind SEO and blog cadence.
  • Incumbent platform work (Shopify) stayed in data-model and compliance plumbing while smaller players chased the agent interface.
  • Fulfillment and inventory operators kept sanding floor-level friction: ShipHero pack-station tools, ShipMonk coverage expansion, inFlow Xero sync, Ordoro receiving.

Watch this week

The agent-interface race is the story to track. Wheelhouse, ShipHero, and Spree all shipped agent-facing surfaces this week, and Wheelhouse's July hackathon and Spree's skills library are explicit attempts to pull third-party builders onto those surfaces — watch whether that seeds visible integrations or stays a demo. The open question is whether the largest platform stays out: Shopify shipped eleven improvements and not one touched an agent interface, so any move it makes toward exposing the admin to assistants would reset expectations for the whole sector. Meanwhile, discount the velocity leaderboard — Printful, Cin7, ShipBob, and the other content-only feeds will keep posting daily with nothing shipping behind it.