Building the intelligence layer for retail and logistics.

A 10:00 pickup. A 4–6 pm slot. A same-day cutoff the day before Mother's Day. Behind each one, someone is making calls, re-slotting orders, messaging drivers and answering "where is it?". That coordination is the hidden cost of local fulfillment. It's manual, it grows with every store you add, and it breaks on the busiest days.

We started by putting the promise, and the work that keeps it, on one record. The next step is to let software take on more of that coordination: carefully, one job at a time, and only where you allow it. We're building toward an operating system for local commerce, and we'll only call each piece done when it ships.

Most of a dispatcher's day is exceptions: a late route, a missed prep deadline, a customer asking where their order is. We want agents to handle those jobs the way a good dispatcher would, inside your rules, with your approval, and with a receipt for everything they do.

Today: at-risk alerts, workflows and customer notifications are live; the AI Agent step in workflows is in early access and read-only.

Next: Exception Agent will propose a fix for an at-risk delivery and act only after you approve.

Later, each a vision: a Promise Agent that recommends opening, closing or re-pricing slots from booking and kept-rate history · a Pickup Agent that sequences preparation against slots and flags orders that won't be ready, once a store-floor app exists · an Update Agent that answers "where is my order?" across help desk, SMS and WhatsApp from live data · a Dispatch Agent that assigns and batches orders under your cost and on-time rules (automated dispatch today is rules-based early access, not AI) · a Proof Agent that checks proof of delivery for gaps.

First: an approval inbox, receipts, undo and a pause switch for every store.

Not every order is local. We want every order, whether it goes by your driver, a courier or a carrier, to be promised and kept from the same Scroll.

Today: labels and route manifests for local orders are live; multi-carrier rates, labels and customs are in early access.

Later, a vision: choosing the right way to send each order and keeping one record either way, shipping from stores, and a courier fallback in the US when your own drivers are full.

First: a confirmed, published set of live carriers.

A network of stores should promise like one business: when one store's slots fill, the next-best store or window is offered, within your rules.

Today: location rules and fulfill-from-matched-location are live; order split and multi-warehouse sourcing are in early access.

Later, a vision: promises balanced across stores automatically, within the rules you set.

Own drivers, couriers and vehicles, planned together, so you can see which mix keeps the most promises at the lowest cost.

Today: the fleet registry, route optimization, the live driver map and driver scorecards are live; automated dispatch, batching and a 3D load-fit check are in early access.

Later, a vision: one plan across your fleet and your courier partners.

A slot is only a real promise if the store has the stock and the time to prepare it.

Today: per-location availability rules that warn about, or block, a pickup where the item isn't stocked are live; order split is in early access.

Later, a vision: a slot offered only when the store has the item and the time to prepare it.

Today you see a late route when its live ETA slips past the window. We want to use the history in the Scroll to show which slots, stores and routes will come under pressure, early enough to act.

Today: live, traffic-aware ETAs and at-risk alerts are live, and so are driver scorecards, operations analytics and reports. These count what happened; nothing here predicts.

Later, a vision: early warnings built from your own promise history.

We'll describe how it works, and how accurate it is, before we call it a feature.

Local promises start in more places than one storefront, and increasingly in AI assistants. We want any channel to ask ScrollEngine for a slot, and any assistant to ask where an order is.

Today: ScrollEngine integrates with Shopify, Adobe Commerce, WooCommerce, BigCommerce and other platforms, and CSV import, API access and signed webhooks are live.

Later: a public delivery-promise API for any storefront, and a read-only MCP server that lets AI assistants ask about slots, orders and ETAs, each with its status in the roadmap below.

First: authentication, rate limits and a published API reference.

Every place that prepares a local order, whether a store, a kitchen, a dark store or a warehouse, should work from the same list of what's due and when, so the promise made at checkout is the one the floor prepares for.

Today: the preparation planner and delivery calendar are live; order split and multi-warehouse sourcing are in early access.

Later, a vision: a store-floor app for picking and handoff.

First: picking and handoff steps recorded on the Scroll, per order.