A platform you run your business on should tell you where it is going. The board below is the real order of work, and the log underneath it is every release since — fixes included.
What is being built, what is queued behind it, and what we are still thinking about. Shipped work moves to the log.
In active development. Dates slip, but these are the things being built this quarter.
The web POS works, but it assumes a connection and a keyboard. The native app is built for a counter: card reader pairing, receipt printing, and a local cache that keeps selling through a dropped Wi-Fi connection, then reconciles the queue when it comes back.
Same catalog, same stock pool, same customer record as the rest of the hub. A shop sale decrements the stock a marketplace listing is quoting from, in the same second.
FBA inventory currently syncs as a single available number. That is enough to avoid overselling and not enough to run a business on — it hides units in transit, units held for inspection, and the ones Amazon has quietly lost.
The rebuild models FBA as a fulfilment location with its own ledger: inbound shipments, reserved units, unfulfillable stock, removals, and a reconciliation report that flags the gaps worth opening a reimbursement case over.
One stock pool is the right default and the wrong answer for anyone running a warehouse alongside three shops. Allocation rules let each channel draw from a named set of locations, with buffers so a marketplace listing never eats the last unit off a shop floor.
Rules are evaluated at read time, so changing a buffer republishes availability everywhere without a re-sync.
Specced and queued. Work starts as soon as something above it ships.
Subscriptions are the clearest case of the app-tax problem: a recurring order is a first-class commerce object, and paying a percentage of it to a third party to schedule a charge is hard to justify.
Native recurring orders, running through whichever processor you already use, with dunning, customer-facing pauses, plan changes, and the resulting orders landing in the same queue as everything else.
Automatic conversion produces prices no merchandiser would choose and margins nobody can predict. Price books get explicit per-currency amounts, with rounding rules and an optional rate-based suggestion you can override.
On the settlement side, orders keep both the presentment amount and the amount that landed in your account, so the payout reconciliation stops being a spreadsheet exercise.
The marketing suite can already send to a list. What it cannot do yet is define that list from behaviour — bought twice in ninety days, only ever through the marketplace, lapsed since spring.
Segments become a shared object: the same definition drives a campaign, gates a promotion, and picks the price list a B2B account sees.
Problems we think are worth solving. No commitment, and no date — yet.
The API is public and documented, which is not the same as having somewhere to start. A pair of maintained starter storefronts would cut the first week of a headless build down to an afternoon.
The open question is the maintenance contract. A starter kit that drifts a year behind the API is worse than no starter kit, so this only ships if we can commit to keeping it current.
Every marketplace wants the same product described its own way — its taxonomy, its attribute names, its title length limits. Doing that by hand is why catalogs go stale on the third channel.
The version worth building drafts the mapping and holds it for review. Nothing reaches a live listing without someone approving it, and every generated field stays traceable back to the master record.
Agencies already run client accounts under one organisation. The part they still do in a spreadsheet is rebilling — taking platform costs, adding their retainer, and invoicing under their own name.
Doing this properly means tax handling across jurisdictions and a clear answer to who owns the customer relationship when an account moves on. Both are worth getting right before writing code.
Everything that has shipped, newest first — including the fixes that never make it into a launch post.
Until now an order was fulfilled from wherever you told it to be fulfilled from, one location per order. That falls apart the moment a basket mixes a warehouse item with something only the Lisbon shop stocks.
B2B and B2C on one catalog was already the point. This release adds the pricing machinery that makes it usable for wholesale accounts.
Promotions were a single discount code with a percentage on it. Anything beyond that meant an app and a monthly bill.
The order things get built in comes from the merchants running on the platform.
Tell us what is blocking you and we will tell you honestly where it sits.