The whole batch. One chat.
Colts 1.4 finishes the job in InventoryLab. Operator now reads every one of your reports — so it can actually answer "what hasn't sold?" — and it builds a batch end to end, prep owner and all, so the flow you start in chat runs all the way to placement.
ChatGPT and Claude can't see your InventoryLab data or build a shipment for you. Operator reads your reports and runs the whole batch — so instead of answering questions, it does the work.
Operator now reads all of your InventoryLab reports — so when you ask "what hasn't sold in 30 days?" or "which items need a restock?", it actually has the data to answer. The questions that used to come back empty now come back with the list.
The batch flow now completes — including setting the prep owner, the step that used to stop it short. Tell Operator what you're shipping and it goes all the way to placement.
The question that used to come back empty.
Before, asking Operator about your InventoryLab data could hit a wall — it wasn't connected to every report, so a simple question like "what hasn't sold in 15 days?" came back with nothing. In 1.4 it reads all of your reports. The same question now comes back with the actual list, ready to act on.
"Which items haven't sold in 15 days?"
No data available.
Couldn't reach the report
"Which items haven't sold in 15 days?"
14 SKUs, oldest first — with days since last sale.
Reads the report and answers
A batch, start to finish.
Building a batch in chat used to stop partway — it couldn't set the prep owner, so the flow stalled before it finished. In 1.4 that step works, and the whole batch runs to completion: products, boxing, prep owner, labels, and placement, without leaving the chat.
Quicker, again.
Reading reports and building a full batch means more steps behind the scenes — so 1.4 keeps tightening the work underneath. Less overhead per call, faster handoffs between steps, so even the longer jobs feel quick.
New prompts to try.
A few things 1.4 makes possible — now that Operator can read every report and finish a batch.
Which of my items haven't sold in the last 30 days since I last replenished them?
Based on my restock report, which products should I reorder this week, and how many of each?
Create a batch for {insert MSKU} — 30 units, $2 cost, one 12×12×12 box at 10 lb — set me as the prep owner and show placement.
Pull my returns from the last 60 days and tell me which products are getting sent back the most.
Shipped, daily.
Alongside the headline features, small fixes land every few days. Below is the full history — everything is already live in AI Operator, nothing to install.
Wholesale, Excel & smart alerts. Operator started running wholesale searches, updating inventory from XLSX files, and scoping FeedbackWhiz alerts.
Matching, repricing & reverse search. Operator started catching bad matches, changing min/max prices, and finding sources from a list of ASINs.
20× faster. Every action sped up, and SmartRepricer and FeedbackWhiz came online — more to ask, less waiting.
The first release. Tactical Arbitrage Quick Picks and product search, and creating an InventoryLab batch — all from chat.
Give it a store and a category — Operator finds the right category page on its own and suggests search phrases to run. No more hunting for category URLs.
Answers come back in seconds instead of minutes, and now cover everything SmartRepricer can do — it no longer refuses tasks it can actually handle.
Quality checks now run daily — regressions get caught and fixed before you ever feel them.
Reliability groundwork — new internal testing checks every change against real prompts before it ships.
Operator creates recipient lists correctly now, and is clear that product lists are a snapshot, not an auto-updating feed.
More dependable Tactical Arbitrage runs — tighter behind-the-scenes tuning so searches behave more consistently.
Update your inventory by uploading an Excel (XLSX) file — spreadsheets are supported now, not just CSV.
When you upload a spreadsheet of leads to verify, Operator now spots the Amazon link column on its own and ignores image columns — fewer setup questions.
Sharper answers on return reasons and listing monitoring.
Fewer dropped connections during long tasks — sessions hold steady through brief busy spells instead of cutting out.
Reverse search got more reliable — Operator finds similar listings for an ASIN and keeps you posted on long Quick Picks jobs without you refreshing.
Adding lots of leads to a collection at once is now much faster.
Operator can now answer SmartRepricer questions end to end — sign-in and errors, products and pricing, settings, and workflows.
Brief connection hiccups when a chat is starting up now recover on their own instead of surfacing as an error.
Chats no longer freeze on long tasks — the connection stays alive while Operator works.
Operator can now handle FeedbackWhiz in bulk — attaching labels, updating status, and exports across many products at once.
More reliable product search and Quick Picks queries.
Submit finds straight to Tactical Edge, and hover an image link in a results spreadsheet to preview the picture.
In dark mode, the fade at the bottom of the chat now blends with the background instead of showing as a gray band over your last messages.
Answers about fulfillment and product labels are now accurate.
Quick Picks searches now honor the Standard vs. Oversize size filter.
Spreadsheet previews now show product images as thumbnails instead of long links.
Answers brought fully up to date with the latest FeedbackWhiz features.
The Return key now starts a new line instead of sending your message, and the scroll-to-bottom button no longer covers the chat box or your last message.
Drag and drop files anywhere on the chat to attach them — no need to reach for the + button.
Spreadsheet exports now include only the columns you asked for, with cleaner defaults — and the bug where a downloaded file came back empty is fixed.
Operator now only suggests apps you actually have, instead of offering features you can't use.
Fixed an error that could make every message fail, and false "FeedbackWhiz is down" timeouts that were never FeedbackWhiz's fault.
Chats now name themselves properly instead of staying stuck on "New Chat" — and if your connection drops mid-task, the chat reconnects and picks up where it left off instead of resetting.
Long tasks now run to completion instead of stopping around the five-minute mark, and busy periods cause far fewer random errors — with five times more room for simultaneous chats.
Creating alerts works again, and Operator can now answer FeedbackWhiz questions for Walmart as well as Amazon.
PDF invoice analysis works again.
The flow you started,
now it finishes.
No setup, nothing to learn. The InventoryLab questions that came back empty now come back answered — and the batch you start in chat runs all the way to placement, with your okay at every write.